UNIVERSIDAD PABLO DE OLAVIDE
Publications
2024
M. J. Jiménez-Navarro and A.R. Troncoso-García and A. Troncoso and F. Martínez-Álvarez and M. Martínez-Ballesteros
Explainable Deep Learning with Embedded Feature Selection for Electricity Demand Forecasting Conference
SST International Conference on Smart Systems and Technologies, 2024.
Abstract | BibTeX | Tags: deep learning, energy, feature selection, XAI
@conference{SST2024,
title = {Explainable Deep Learning with Embedded Feature Selection for Electricity Demand Forecasting},
author = {M. J. Jiménez-Navarro and A.R. Troncoso-García and A. Troncoso and F. Martínez-Álvarez and M. Martínez-Ballesteros},
year = {2024},
date = {2024-10-16},
urldate = {2024-10-16},
booktitle = {SST International Conference on Smart Systems and Technologies},
abstract = {Electricity demand forecasting is an important part of the energy industry strategy. Accurate predictions are crucial for maintaining a stable energy supply, planning production, managing distribution, preventing grid overloads, integrating renewable energy sources, and reducing costs and environmental impact. Machine learning and, in particular, deep learning are promising techniques to improve the prediction accuracy of electric demand, but face challenges related to a lack of interpretability due to the “black box” nature of some models. Feature selection methods address these issues by identifying relevant features and simplifying the learning process. This paper aims to explain the most critical lags that impact electric demand forecasting in Spain using the temporal selection layer technique within deep learning models for time series forecasting. This technique transforms a neural network into a model with embedded feature selection, aiming to enhance efficacy and interpretability while reducing computational costs. The results were compared with other methods that incorporate an embedded feature selection mechanism to select the best model. Furthermore, an explainable technique is used to assess the feature importance in the best model over the last year to understand how input features influence electric demand forecasting and provide insights into their contributions and interactions. The results show that our approach improves both the efficacy and interpretability in the context of electric demand forecasting.},
keywords = {deep learning, energy, feature selection, XAI},
pubstate = {published},
tppubtype = {conference}
}
Electricity demand forecasting is an important part of the energy industry strategy. Accurate predictions are crucial for maintaining a stable energy supply, planning production, managing distribution, preventing grid overloads, integrating renewable energy sources, and reducing costs and environmental impact. Machine learning and, in particular, deep learning are promising techniques to improve the prediction accuracy of electric demand, but face challenges related to a lack of interpretability due to the “black box” nature of some models. Feature selection methods address these issues by identifying relevant features and simplifying the learning process. This paper aims to explain the most critical lags that impact electric demand forecasting in Spain using the temporal selection layer technique within deep learning models for time series forecasting. This technique transforms a neural network into a model with embedded feature selection, aiming to enhance efficacy and interpretability while reducing computational costs. The results were compared with other methods that incorporate an embedded feature selection mechanism to select the best model. Furthermore, an explainable technique is used to assess the feature importance in the best model over the last year to understand how input features influence electric demand forecasting and provide insights into their contributions and interactions. The results show that our approach improves both the efficacy and interpretability in the context of electric demand forecasting.
M. J. Jiménez-Navarro and M. Martínez-Ballesteros and I. S. Brito and F. Martínez-Álvarez and G. Asencio-Cortés
Embedded feature selection for neural networks via learnable drop layer Journal Article
In: Logic Journal of the IGPL, pp. jzae062, 2024.
Abstract | Links | BibTeX | Tags: deep learning, feature selection, time series
@article{JIMENEZ-NAVARRO24b,
title = {Embedded feature selection for neural networks via learnable drop layer},
author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and I. S. Brito and F. Martínez-Álvarez and G. Asencio-Cortés},
url = {https://academic.oup.com/jigpal/advance-article/doi/10.1093/jigpal/jzae062/7689640},
doi = {https://doi.org/10.1093/jigpal/jzae062},
year = {2024},
date = {2024-07-06},
urldate = {2024-07-06},
journal = {Logic Journal of the IGPL},
pages = {jzae062},
abstract = {Feature selection is a widely studied technique whose goal is to reduce the dimensionality of the problem by removing irrelevant features. It has multiple benefits, such as improved efficacy, efficiency and interpretability of almost any type of machine learning model. Feature selection techniques may be divided into three main categories, depending on the process used to remove the features known as Filter, Wrapper and Embedded. Embedded methods are usually the preferred feature selection method that efficiently obtains a selection of the most relevant features of the model. However, not all models support an embedded feature selection that forces the use of a different method, reducing the efficiency and reliability of the selection. Neural networks are an example of a model that does not support embedded feature selection. As neural networks have shown to provide remarkable results in multiple scenarios such as classification and regression, sometimes in an ensemble with a model that includes an embedded feature selection, we attempt to embed a feature selection process with a general-purpose methodology. In this work, we propose a novel general-purpose layer for neural networks that removes the influence of irrelevant features. The Feature-Aware Drop Layer is included at the top of the neural network and trained during the backpropagation process without any additional parameters. Our methodology is tested with 17 datasets for classification and regression tasks, including data from different fields such as Health, Economic and Environment, among others. The results show remarkable improvements compared to three different feature selection approaches, with reliable, efficient and effective results.},
keywords = {deep learning, feature selection, time series},
pubstate = {published},
tppubtype = {article}
}
Feature selection is a widely studied technique whose goal is to reduce the dimensionality of the problem by removing irrelevant features. It has multiple benefits, such as improved efficacy, efficiency and interpretability of almost any type of machine learning model. Feature selection techniques may be divided into three main categories, depending on the process used to remove the features known as Filter, Wrapper and Embedded. Embedded methods are usually the preferred feature selection method that efficiently obtains a selection of the most relevant features of the model. However, not all models support an embedded feature selection that forces the use of a different method, reducing the efficiency and reliability of the selection. Neural networks are an example of a model that does not support embedded feature selection. As neural networks have shown to provide remarkable results in multiple scenarios such as classification and regression, sometimes in an ensemble with a model that includes an embedded feature selection, we attempt to embed a feature selection process with a general-purpose methodology. In this work, we propose a novel general-purpose layer for neural networks that removes the influence of irrelevant features. The Feature-Aware Drop Layer is included at the top of the neural network and trained during the backpropagation process without any additional parameters. Our methodology is tested with 17 datasets for classification and regression tasks, including data from different fields such as Health, Economic and Environment, among others. The results show remarkable improvements compared to three different feature selection approaches, with reliable, efficient and effective results.
F. Rodríguez-Díaz and J. F. Torres and D. Gutiérrez-Avilés and A. Troncoso and F. Martínez-Álvarez
CAEPIA Conference of the Spanish Association for Artificial Intelligence, Lecture Notes in Artificial Intelligence 2024.
Abstract | Links | BibTeX | Tags: quantum computing
@conference{RODRIGUEZ-DIAZ24b,
title = {An Experimental Comparison of Qiskit and Pennylane for Hybrid Quantum-Classical Support Vector Machines},
author = {F. Rodríguez-Díaz and J. F. Torres and D. Gutiérrez-Avilés and A. Troncoso and F. Martínez-Álvarez},
url = {https://link.springer.com/chapter/10.1007/978-3-031-62799-6_13},
doi = {https://doi.org/10.1007/978-3-031-62799-6_13},
year = {2024},
date = {2024-06-06},
booktitle = {CAEPIA Conference of the Spanish Association for Artificial Intelligence},
pages = {121-130},
series = {Lecture Notes in Artificial Intelligence},
abstract = {Quantum computing holds great promise for enhancing machine learning algorithms, particularly by integrating classical and quantum techniques. This study compares two prominent quantum development frameworks, Qiskit and Pennylane, focusing on their suitability for hybrid quantum-classical support vector machines with quantum kernels. Our analysis reveals that Qiskit requires less theoretical information to be used, while Pennylane demonstrates superior performance in terms of execution time. Although both frameworks exhibit variances, our experiments reveal that Qiskit consistently yields superior classification accuracy compared to Pennylane when training classifiers with quantum kernels. Additionally, our results suggest that the performance of both frameworks remains stable for up to 20 qubits, indicating their suitability for practical applications. Overall, our findings provide valuable insights into the strengths and limitations of Qiskit and Pennylane for hybrid quantum-classical machine learning.},
keywords = {quantum computing},
pubstate = {published},
tppubtype = {conference}
}
Quantum computing holds great promise for enhancing machine learning algorithms, particularly by integrating classical and quantum techniques. This study compares two prominent quantum development frameworks, Qiskit and Pennylane, focusing on their suitability for hybrid quantum-classical support vector machines with quantum kernels. Our analysis reveals that Qiskit requires less theoretical information to be used, while Pennylane demonstrates superior performance in terms of execution time. Although both frameworks exhibit variances, our experiments reveal that Qiskit consistently yields superior classification accuracy compared to Pennylane when training classifiers with quantum kernels. Additionally, our results suggest that the performance of both frameworks remains stable for up to 20 qubits, indicating their suitability for practical applications. Overall, our findings provide valuable insights into the strengths and limitations of Qiskit and Pennylane for hybrid quantum-classical machine learning.
A. R. Troncoso-García and M. J. Jiménez-Navarro and F. Martínez-Álvarez and A. Troncoso
Ground-Level Ozone Forecasting using Explainable Machine Learning Conference
CAEPIA Conference of the Spanish Association for Artificial Intelligence, vol. 14640, Lecture Notes in Artificial Intelligence 2024.
Abstract | Links | BibTeX | Tags: forecasting, time series, XAI
@conference{TRONCOSO-GARCIA24c,
title = {Ground-Level Ozone Forecasting using Explainable Machine Learning},
author = {A. R. Troncoso-García and M. J. Jiménez-Navarro and F. Martínez-Álvarez and A. Troncoso},
url = {https://link.springer.com/chapter/10.1007/978-3-031-62799-6_8},
doi = {https://doi.org/10.1007/978-3-031-62799-6_8},
year = {2024},
date = {2024-06-06},
booktitle = {CAEPIA Conference of the Spanish Association for Artificial Intelligence},
volume = {14640},
pages = {71-80},
series = {Lecture Notes in Artificial Intelligence},
abstract = {The ozone concentration at ground level is a pivotal indicator of air quality, as elevated ozone levels can lead to adverse effects on the environment. In this study various machine learning models for ground-level ozone forecasting are optimised using a Bayesian technique. Predictions are obtained 24 h in advance using historical ozone data and related environmental variables, including meteorological measurements and other air quality indicators. The results indicated that the Extra Trees model emerges as the optimal solution, showcasing competitive performance alongside reasonable training times. Furthermore, an explainable artificial intelligence technique is applied to enhance the interpretability of model predictions, providing insights into the contribution of input features to the predictions computed by the model. The features identified as important, namely PM10, air temperature and CO2 concentration, are validated as key factors in the literature to forecast ground-level ozone concentration.},
keywords = {forecasting, time series, XAI},
pubstate = {published},
tppubtype = {conference}
}
The ozone concentration at ground level is a pivotal indicator of air quality, as elevated ozone levels can lead to adverse effects on the environment. In this study various machine learning models for ground-level ozone forecasting are optimised using a Bayesian technique. Predictions are obtained 24 h in advance using historical ozone data and related environmental variables, including meteorological measurements and other air quality indicators. The results indicated that the Extra Trees model emerges as the optimal solution, showcasing competitive performance alongside reasonable training times. Furthermore, an explainable artificial intelligence technique is applied to enhance the interpretability of model predictions, providing insights into the contribution of input features to the predictions computed by the model. The features identified as important, namely PM10, air temperature and CO2 concentration, are validated as key factors in the literature to forecast ground-level ozone concentration.
M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Martínez-Álvarez and G. Asencio-Cortés
Explaining deep learning models for ozone pollution prediction via embedded feature selection Journal Article
In: Applied Soft Computing, vol. 157, pp. 111504, 2024.
Abstract | Links | BibTeX | Tags: deep learning, feature selection, time series, XAI
@article{JIMENEZ-NAVARRO24,
title = {Explaining deep learning models for ozone pollution prediction via embedded feature selection},
author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Martínez-Álvarez and G. Asencio-Cortés},
url = {https://www.sciencedirect.com/science/article/pii/S1568494624002783},
doi = {https://doi.org/10.1016/j.asoc.2024.111504},
year = {2024},
date = {2024-04-04},
journal = {Applied Soft Computing},
volume = {157},
pages = {111504},
abstract = {Ambient air pollution is a pervasive global issue that poses significant health risks. Among pollutants, ozone (O3) is responsible for an estimated 1 to 1.2 million premature deaths yearly. Furthermore, O3 adversely affects climate warming, crop productivity, and more. Its formation occurs when nitrogen oxides and volatile organic compounds react with short-wavelength solar radiation. Consequently, urban areas with high traffic volume and elevated temperatures are particularly prone to elevated O3 levels, which pose a significant health risk to their inhabitants. In response to this problem, many countries have developed web and mobile applications that provide real-time air pollution information using sensor data. However, while these applications offer valuable insight into current pollution levels, predicting future pollutant behavior is crucial for effective planning and mitigation strategies. Therefore, our main objectives are to develop accurate and efficient prediction models and identify the key factors that influence O3 levels. We adopt a time series forecasting approach to address these objectives, which allows us to analyze and predict O3 future behavior. Additionally, we tackle the feature selection problem to identify the most relevant features and periods that contribute to prediction accuracy by introducing a novel method called the Time Selection Layer in Deep Learning models, which significantly improves model performance, reduces complexity, and enhances interpretability. Our study focuses on data collected from five representative areas in Seville, Cordova, and Jaen provinces in Spain, using multiple sensors to capture comprehensive pollution data. We compare the performance of three models: Lasso, Decision Tree, and Deep Learning with and without incorporating the Time Selection Layer. Our results demonstrate that including the Time Selection Layer significantly enhances the effectiveness and interpretability of Deep Learning models, achieving an average effectiveness improvement of 9% across all monitored areas.},
keywords = {deep learning, feature selection, time series, XAI},
pubstate = {published},
tppubtype = {article}
}
Ambient air pollution is a pervasive global issue that poses significant health risks. Among pollutants, ozone (O3) is responsible for an estimated 1 to 1.2 million premature deaths yearly. Furthermore, O3 adversely affects climate warming, crop productivity, and more. Its formation occurs when nitrogen oxides and volatile organic compounds react with short-wavelength solar radiation. Consequently, urban areas with high traffic volume and elevated temperatures are particularly prone to elevated O3 levels, which pose a significant health risk to their inhabitants. In response to this problem, many countries have developed web and mobile applications that provide real-time air pollution information using sensor data. However, while these applications offer valuable insight into current pollution levels, predicting future pollutant behavior is crucial for effective planning and mitigation strategies. Therefore, our main objectives are to develop accurate and efficient prediction models and identify the key factors that influence O3 levels. We adopt a time series forecasting approach to address these objectives, which allows us to analyze and predict O3 future behavior. Additionally, we tackle the feature selection problem to identify the most relevant features and periods that contribute to prediction accuracy by introducing a novel method called the Time Selection Layer in Deep Learning models, which significantly improves model performance, reduces complexity, and enhances interpretability. Our study focuses on data collected from five representative areas in Seville, Cordova, and Jaen provinces in Spain, using multiple sensors to capture comprehensive pollution data. We compare the performance of three models: Lasso, Decision Tree, and Deep Learning with and without incorporating the Time Selection Layer. Our results demonstrate that including the Time Selection Layer significantly enhances the effectiveness and interpretability of Deep Learning models, achieving an average effectiveness improvement of 9% across all monitored areas.
D. Gutiérrez-Avilés and J. F. Torres and F. Martínez-Álvarez and J. Cugliari
An evolutionary triclustering approach to discover electricity consumption patterns in France Conference
SAC 39th Annual ACM Symposium on Applied Computing, 2024.
Abstract | BibTeX | Tags: clustering, energy, time series
@conference{GUTIERREZ24_SAC,
title = {An evolutionary triclustering approach to discover electricity consumption patterns in France},
author = {D. Gutiérrez-Avilés and J. F. Torres and F. Martínez-Álvarez and J. Cugliari},
year = {2024},
date = {2024-02-04},
booktitle = {SAC 39th Annual ACM Symposium on Applied Computing},
pages = {386-394},
abstract = {Electricity consumption patterns are critical in shaping energy policies
and optimizing resource allocation. In pursuing a more sustainable
and efficient energy future, uncovering hidden consumption
patterns is paramount. This paper introduces an innovative approach,
leveraging evolutionary triclustering techniques, to unveil
previously undisclosed electricity consumption patterns in France.
By harnessing the power of triclustering algorithms, this research
provides a comprehensive analysis of electricity usage across various
dimensions, shedding light on intricate relationships among
variables. Using this novel method, the study reveals concealed
patterns and offers insights that can inform decision-makers and
stakeholders in the energy sector. The findings contribute to a better
understanding of electricity consumption dynamics, aiding in
developing more targeted and effective energy management strategies.
This research represents a significant step forward in the quest
for sustainable energy solutions and underscores the potential of
evolutionary triclustering as a valuable tool in uncovering complex
consumption patterns.},
keywords = {clustering, energy, time series},
pubstate = {published},
tppubtype = {conference}
}
Electricity consumption patterns are critical in shaping energy policies
and optimizing resource allocation. In pursuing a more sustainable
and efficient energy future, uncovering hidden consumption
patterns is paramount. This paper introduces an innovative approach,
leveraging evolutionary triclustering techniques, to unveil
previously undisclosed electricity consumption patterns in France.
By harnessing the power of triclustering algorithms, this research
provides a comprehensive analysis of electricity usage across various
dimensions, shedding light on intricate relationships among
variables. Using this novel method, the study reveals concealed
patterns and offers insights that can inform decision-makers and
stakeholders in the energy sector. The findings contribute to a better
understanding of electricity consumption dynamics, aiding in
developing more targeted and effective energy management strategies.
This research represents a significant step forward in the quest
for sustainable energy solutions and underscores the potential of
evolutionary triclustering as a valuable tool in uncovering complex
consumption patterns.
and optimizing resource allocation. In pursuing a more sustainable
and efficient energy future, uncovering hidden consumption
patterns is paramount. This paper introduces an innovative approach,
leveraging evolutionary triclustering techniques, to unveil
previously undisclosed electricity consumption patterns in France.
By harnessing the power of triclustering algorithms, this research
provides a comprehensive analysis of electricity usage across various
dimensions, shedding light on intricate relationships among
variables. Using this novel method, the study reveals concealed
patterns and offers insights that can inform decision-makers and
stakeholders in the energy sector. The findings contribute to a better
understanding of electricity consumption dynamics, aiding in
developing more targeted and effective energy management strategies.
This research represents a significant step forward in the quest
for sustainable energy solutions and underscores the potential of
evolutionary triclustering as a valuable tool in uncovering complex
consumption patterns.
R. Pérez-Chacón and G. Asencio-Cortés and A. Troncoso and F. Martínez-Álvarez
Pattern sequence-based algorithm for multivariate big data time series forecasting: Application to electricity consumption Journal Article
In: Future Generation Computer Systems, vol. 154, pp. 397-412, 2024.
Abstract | Links | BibTeX | Tags: big data, energy, forecasting, time series
@article{PEREZ24,
title = {Pattern sequence-based algorithm for multivariate big data time series forecasting: Application to electricity consumption},
author = {R. Pérez-Chacón and G. Asencio-Cortés and A. Troncoso and F. Martínez-Álvarez},
url = {https://www.sciencedirect.com/science/article/pii/S0167739X23004752},
doi = {https://doi.org/10.1016/j.future.2023.12.021},
year = {2024},
date = {2024-01-29},
journal = {Future Generation Computer Systems},
volume = {154},
pages = {397-412},
abstract = {Several interrelated variables typically characterize real-world processes, and a time series cannot be predicted without considering the influence that other time series might have on the target time series. This work proposes a novel algorithm to forecast multivariate big data time series. This new general-purpose approach consists first of a previous pattern recognition performed jointly using all time series that form the multivariate time series and then predicts the target time series by searching for similarities between pattern sequences. The proposed algorithm is designed to tackle multivariate time series forecasting problems within the context of big data. In particular, the algorithm has been developed with a distributed nature to enhance its efficiency in analyzing and processing large volumes of data. Moreover, the algorithm is straightforward to use, with only two parameters needing adjustment. Another advantage of the MV-bigPSF algorithm is its ability to perform multi-step forecasting, which is particularly useful in many practical applications. To evaluate the algorithm’s performance, real-world data from Uruguay’s power consumption has been utilized. Specifically, MV-bigPSF has been compared with both univariate and multivariate methods. Regarding the univariate ones, MV-bigPSF improved 12.8% in MAPE compared to the second-best method. Regarding the multivariate comparison, MV-bigPSF improved 44.8% in MAPE with respect to the second most accurate method. Regarding efficiency, the execution time of MV-bigPSF was 1.83 times faster than the second-fastest multivariate method, both in a single-core environment. Therefore, the proposed algorithm can be a valuable tool for practitioners and researchers working in multivariate time series forecasting, particularly in big data applications.},
keywords = {big data, energy, forecasting, time series},
pubstate = {published},
tppubtype = {article}
}
Several interrelated variables typically characterize real-world processes, and a time series cannot be predicted without considering the influence that other time series might have on the target time series. This work proposes a novel algorithm to forecast multivariate big data time series. This new general-purpose approach consists first of a previous pattern recognition performed jointly using all time series that form the multivariate time series and then predicts the target time series by searching for similarities between pattern sequences. The proposed algorithm is designed to tackle multivariate time series forecasting problems within the context of big data. In particular, the algorithm has been developed with a distributed nature to enhance its efficiency in analyzing and processing large volumes of data. Moreover, the algorithm is straightforward to use, with only two parameters needing adjustment. Another advantage of the MV-bigPSF algorithm is its ability to perform multi-step forecasting, which is particularly useful in many practical applications. To evaluate the algorithm’s performance, real-world data from Uruguay’s power consumption has been utilized. Specifically, MV-bigPSF has been compared with both univariate and multivariate methods. Regarding the univariate ones, MV-bigPSF improved 12.8% in MAPE compared to the second-best method. Regarding the multivariate comparison, MV-bigPSF improved 44.8% in MAPE with respect to the second most accurate method. Regarding efficiency, the execution time of MV-bigPSF was 1.83 times faster than the second-fastest multivariate method, both in a single-core environment. Therefore, the proposed algorithm can be a valuable tool for practitioners and researchers working in multivariate time series forecasting, particularly in big data applications.
C. G. García-Soto and J. F. Torres and M. A. Zamora-Izquierdo and J. Palma and A. Troncoso
Water consumption time series forecasting in urban centers using deep neural networks Journal Article
In: Applied Water Science, vol. 14, pp. 1-14, 2024.
Abstract | Links | BibTeX | Tags: deep learning, forecasting, time series
@article{GARCIA-SOTO24,
title = {Water consumption time series forecasting in urban centers using deep neural networks},
author = {C. G. García-Soto and J. F. Torres and M. A. Zamora-Izquierdo and J. Palma and A. Troncoso},
url = {https://link.springer.com/article/10.1007/s13201-023-02072-4},
doi = {https://doi.org/10.1007/s13201-023-02072-4},
year = {2024},
date = {2024-01-12},
journal = {Applied Water Science},
volume = {14},
pages = {1-14},
abstract = {The time series analysis and prediction techniques are highly valued in many application felds, such as economy, medicine and biology, environmental sciences or meteorology, among others. In the last years, there is a growing interest in the sustainable and optimal management of a resource as scarce as essential: the water. Forecasting techniques for water management can be used for diferent time horizons from the planning of constructions that can respond to long-term needs, to the detection of anomalies in the operation of facilities or the optimization of the operation in the short and medium term. In this paper, a deep neural network is specifcally designed to predict water consumption in the short-term. Results are reported using the time series of water consumption for a year and a half measured with 10-min frequency in the city of Murcia, the seventh largest city in Spain by number of inhabitants. The results are compared with K Nearest Neighbors, Random Forest, Extreme Gradient Boosting, Seasonal Autoregressive Integrated Moving Average and two persistence models as naive methods, showing the proposed deep learning model the most accurate results.},
keywords = {deep learning, forecasting, time series},
pubstate = {published},
tppubtype = {article}
}
The time series analysis and prediction techniques are highly valued in many application felds, such as economy, medicine and biology, environmental sciences or meteorology, among others. In the last years, there is a growing interest in the sustainable and optimal management of a resource as scarce as essential: the water. Forecasting techniques for water management can be used for diferent time horizons from the planning of constructions that can respond to long-term needs, to the detection of anomalies in the operation of facilities or the optimization of the operation in the short and medium term. In this paper, a deep neural network is specifcally designed to predict water consumption in the short-term. Results are reported using the time series of water consumption for a year and a half measured with 10-min frequency in the city of Murcia, the seventh largest city in Spain by number of inhabitants. The results are compared with K Nearest Neighbors, Random Forest, Extreme Gradient Boosting, Seasonal Autoregressive Integrated Moving Average and two persistence models as naive methods, showing the proposed deep learning model the most accurate results.
F. Martínez-Álvarez and R. Scitovski and C. Rubio-Escudero and A. Morales-Esteban
Emerging trends in big data analytics and natural disasters (Editorial) Journal Article
In: Computers and Geosciences, vol. 182, pp. 105465, 2024.
Links | BibTeX | Tags: big data, natural disasters, time series
@article{MARTINEZ24,
title = {Emerging trends in big data analytics and natural disasters (Editorial)},
author = {F. Martínez-Álvarez and R. Scitovski and C. Rubio-Escudero and A. Morales-Esteban},
url = {https://www.sciencedirect.com/science/article/pii/S0098300423001693},
doi = {https://doi.org/10.1016/j.cageo.2023.105465},
year = {2024},
date = {2024-01-01},
journal = {Computers and Geosciences},
volume = {182},
pages = {105465},
keywords = {big data, natural disasters, time series},
pubstate = {published},
tppubtype = {article}
}
M. García-Torres and D. P. Pinto-Roa and C. Núñez-Castillo and B. Quiñonez and G. Vázquez and M. Allegretti and M. E. García-Diaz
Feature selection applied to QoS/QoE modeling on video and web-based mobile data services: An ordinal approach Journal Article
In: Computer Communications, 2024.
Abstract | Links | BibTeX | Tags: feature selection
@article{garcia2024feature,
title = {Feature selection applied to QoS/QoE modeling on video and web-based mobile data services: An ordinal approach},
author = {M. García-Torres and D. P. Pinto-Roa and C. Núñez-Castillo and B. Quiñonez and G. Vázquez and M. Allegretti and M. E. García-Diaz},
url = {https://www.sciencedirect.com/science/article/pii/S0140366424000410},
doi = {10.1016/j.comcom.2024.02.004},
year = {2024},
date = {2024-01-01},
journal = {Computer Communications},
publisher = {Elsevier},
abstract = {Nowadays, mobile service providers perceive the user experience as a reliable indicator of the quality associated to a service. Given a set of Quality of Service (QoS) factors, the aim is to predict the Quality of Experience (QoE), measured in terms of the Mean Opinion Score (MOS). Although this problem is receiving much attention, there are still some challenges that require more research in order to find effective solutions for meeting user’s expectation in terms of service quality. A core challenge in this topic refers to the analysis of the contribution of each factor to the QoS/QoE Model. In this work, we study the mapping between QoS and QoE on video and web-based services using a machine learning approach. For such purpose, we design a lab-testing methodology to emulate different cellular transmission network scenarios. Then, we address the problem of inducing a predictive model and identifying relevant QoS factors. Results suggest that bandwidth is a key factor when analyzing user’s perception of service quality.},
keywords = {feature selection},
pubstate = {published},
tppubtype = {article}
}
Nowadays, mobile service providers perceive the user experience as a reliable indicator of the quality associated to a service. Given a set of Quality of Service (QoS) factors, the aim is to predict the Quality of Experience (QoE), measured in terms of the Mean Opinion Score (MOS). Although this problem is receiving much attention, there are still some challenges that require more research in order to find effective solutions for meeting user’s expectation in terms of service quality. A core challenge in this topic refers to the analysis of the contribution of each factor to the QoS/QoE Model. In this work, we study the mapping between QoS and QoE on video and web-based services using a machine learning approach. For such purpose, we design a lab-testing methodology to emulate different cellular transmission network scenarios. Then, we address the problem of inducing a predictive model and identifying relevant QoS factors. Results suggest that bandwidth is a key factor when analyzing user’s perception of service quality.
F. Morales-Mareco and M. García-Torres and F. Divina and D. H Stalder and C. Sauer
Machine learning for electric energy consumption forecasting: Application to the Paraguayan system Journal Article
In: Logic Journal of the IGPL, pp. jzae035, 2024.
Abstract | Links | BibTeX | Tags: energy
@article{morales2024machine,
title = {Machine learning for electric energy consumption forecasting: Application to the Paraguayan system},
author = { F. Morales-Mareco and M. García-Torres and F. Divina and D. H Stalder and C. Sauer},
url = {https://academic.oup.com/jigpal/advance-article-abstract/doi/10.1093/jigpal/jzae035/7639120},
doi = {10.1093/jigpal/jzae035},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Logic Journal of the IGPL},
pages = {jzae035},
publisher = {Oxford University Press},
abstract = {In this paper we address the problem of short-term
electric energy prediction using a time series forecasting approach
applied to data generated by a Paraguayan electricity distribution
provider. The dataset used in this work contains data collected over a
three-year period. This is the first time that these data have been
used; therefore, a preprocessing phase of the data was also performed.
In particular, we propose a comparative study of various machine
learning and statistical strategies with the objective of predicting the
electric energy consumption for a given prediction horizon, in our case
seven days, using historical data. In this paper we have tested the
effectiveness of the techniques with different historical window sizes.
Specifically, we considered two ensemble strategies, a neural network, a
deep learning technique and linear regression. Moreover, in this study,
we tested whether the inclusion of meteorological data can help achieve
better predictions. In particular, we considered data regarding
temperature, humidity, wind speed and atmospheric pressure registered
during the three-year period of data collection. The results show that,
in general, the deep learning approach obtains the best results and that
such results are obtained when meteorological data are also considered.
Moreover, when meteorological data is used, a smaller historical window
size is required to obtain precise predictions.},
keywords = {energy},
pubstate = {published},
tppubtype = {article}
}
In this paper we address the problem of short-term
electric energy prediction using a time series forecasting approach
applied to data generated by a Paraguayan electricity distribution
provider. The dataset used in this work contains data collected over a
three-year period. This is the first time that these data have been
used; therefore, a preprocessing phase of the data was also performed.
In particular, we propose a comparative study of various machine
learning and statistical strategies with the objective of predicting the
electric energy consumption for a given prediction horizon, in our case
seven days, using historical data. In this paper we have tested the
effectiveness of the techniques with different historical window sizes.
Specifically, we considered two ensemble strategies, a neural network, a
deep learning technique and linear regression. Moreover, in this study,
we tested whether the inclusion of meteorological data can help achieve
better predictions. In particular, we considered data regarding
temperature, humidity, wind speed and atmospheric pressure registered
during the three-year period of data collection. The results show that,
in general, the deep learning approach obtains the best results and that
such results are obtained when meteorological data are also considered.
Moreover, when meteorological data is used, a smaller historical window
size is required to obtain precise predictions.
electric energy prediction using a time series forecasting approach
applied to data generated by a Paraguayan electricity distribution
provider. The dataset used in this work contains data collected over a
three-year period. This is the first time that these data have been
used; therefore, a preprocessing phase of the data was also performed.
In particular, we propose a comparative study of various machine
learning and statistical strategies with the objective of predicting the
electric energy consumption for a given prediction horizon, in our case
seven days, using historical data. In this paper we have tested the
effectiveness of the techniques with different historical window sizes.
Specifically, we considered two ensemble strategies, a neural network, a
deep learning technique and linear regression. Moreover, in this study,
we tested whether the inclusion of meteorological data can help achieve
better predictions. In particular, we considered data regarding
temperature, humidity, wind speed and atmospheric pressure registered
during the three-year period of data collection. The results show that,
in general, the deep learning approach obtains the best results and that
such results are obtained when meteorological data are also considered.
Moreover, when meteorological data is used, a smaller historical window
size is required to obtain precise predictions.
G. Sosa-Cabrera and S. Gómez-Guerrero and M. García-Torres and C. E Schaerer
Feature selection: A perspective on inter-attribute cooperation Journal Article
In: International Journal of Data Science and Analytics, vol. 17, no. 2, pp. 139–151, 2024.
Abstract | Links | BibTeX | Tags: feature selection
@article{sosa2024feature,
title = {Feature selection: A perspective on inter-attribute cooperation},
author = { G. Sosa-Cabrera and S. Gómez-Guerrero and M. García-Torres and C. E Schaerer},
url = {https://link.springer.com/article/10.1007/s41060-023-00439-z},
doi = {10.1007/s41060-023-00439-z},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {International Journal of Data Science and Analytics},
volume = {17},
number = {2},
pages = {139–151},
publisher = {Springer},
abstract = {High-dimensional datasets depict a challenge for learning
tasks in data mining and machine learning. Feature selection is an
effective technique in dealing with dimensionality reduction. It is
often an essential data processing step prior to applying a learning
algorithm. Over the decades, filter feature selection methods have
evolved from simple univariate relevance ranking algorithms to more
sophisticated relevance-redundancy trade-offs and to multivariate
dependencies-based approaches in recent years. This tendency to capture
multivariate dependence aims at obtaining unique information about the
class from the intercooperation among features. This paper presents a
comprehensive survey of the state-of-the-art work on filter feature
selection methods assisted by feature intercooperation, and summarizes
the contributions of different approaches found in the literature.
Furthermore, current issues and challenges are introduced to identify
promising future research and development.},
keywords = {feature selection},
pubstate = {published},
tppubtype = {article}
}
High-dimensional datasets depict a challenge for learning
tasks in data mining and machine learning. Feature selection is an
effective technique in dealing with dimensionality reduction. It is
often an essential data processing step prior to applying a learning
algorithm. Over the decades, filter feature selection methods have
evolved from simple univariate relevance ranking algorithms to more
sophisticated relevance-redundancy trade-offs and to multivariate
dependencies-based approaches in recent years. This tendency to capture
multivariate dependence aims at obtaining unique information about the
class from the intercooperation among features. This paper presents a
comprehensive survey of the state-of-the-art work on filter feature
selection methods assisted by feature intercooperation, and summarizes
the contributions of different approaches found in the literature.
Furthermore, current issues and challenges are introduced to identify
promising future research and development.
tasks in data mining and machine learning. Feature selection is an
effective technique in dealing with dimensionality reduction. It is
often an essential data processing step prior to applying a learning
algorithm. Over the decades, filter feature selection methods have
evolved from simple univariate relevance ranking algorithms to more
sophisticated relevance-redundancy trade-offs and to multivariate
dependencies-based approaches in recent years. This tendency to capture
multivariate dependence aims at obtaining unique information about the
class from the intercooperation among features. This paper presents a
comprehensive survey of the state-of-the-art work on filter feature
selection methods assisted by feature intercooperation, and summarizes
the contributions of different approaches found in the literature.
Furthermore, current issues and challenges are introduced to identify
promising future research and development.
F. Divina and M. García-Torres and F. Gómez-Vela and D. S. Rodriguez-Baena
A stacking ensemble learning for Iberian pigs activity prediction: a time series forecasting approach Journal Article
In: AIMS Mathematics, vol. 9, no. 5, pp. 13358–13384, 2024.
Abstract | Links | BibTeX | Tags: time series
@article{divina2024stacking,
title = {A stacking ensemble learning for Iberian pigs activity prediction: a time series forecasting approach},
author = { F. Divina and M. García-Torres and F. Gómez-Vela and D. S. Rodriguez-Baena},
url = {https://www.aimspress.com/article/doi/10.3934/math.2024652},
doi = {10.3934/math.2024652},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {AIMS Mathematics},
volume = {9},
number = {5},
pages = {13358–13384},
abstract = {Automatic determination of abnormal animal activities can
be helpful for the timely detection of signs of health and welfare
problems. Usually, this problem is addressed as a classification
problem, which typically requires manual annotation of behaviors. This
manual annotation can introduce noise into the data and may not always
be possible. This motivated us to address the problem as a time-series
forecasting problem in which the activity of an animal can be predicted.
In this work, different machine learning techniques were tested to
obtain activity patterns for Iberian pigs. In particular, we propose a
novel stacking ensemble learning approach that combines base learners
with meta-learners to obtain the final predictive model. Results confirm
the superior performance of the proposed method relative to the other
tested strategies. We also explored the possibility of using predictive
models trained on an animal to predict the activity of different animals
on the same farm. As expected, the predictive performance degrades in
this case, but it remains acceptable. The proposed method could be
integrated into a monitoring system that may have the potential to
transform the way farm animals are monitored, improving their health and
welfare conditions, for example, by allowing the early detection of a
possible health problem.},
keywords = {time series},
pubstate = {published},
tppubtype = {article}
}
Automatic determination of abnormal animal activities can
be helpful for the timely detection of signs of health and welfare
problems. Usually, this problem is addressed as a classification
problem, which typically requires manual annotation of behaviors. This
manual annotation can introduce noise into the data and may not always
be possible. This motivated us to address the problem as a time-series
forecasting problem in which the activity of an animal can be predicted.
In this work, different machine learning techniques were tested to
obtain activity patterns for Iberian pigs. In particular, we propose a
novel stacking ensemble learning approach that combines base learners
with meta-learners to obtain the final predictive model. Results confirm
the superior performance of the proposed method relative to the other
tested strategies. We also explored the possibility of using predictive
models trained on an animal to predict the activity of different animals
on the same farm. As expected, the predictive performance degrades in
this case, but it remains acceptable. The proposed method could be
integrated into a monitoring system that may have the potential to
transform the way farm animals are monitored, improving their health and
welfare conditions, for example, by allowing the early detection of a
possible health problem.
be helpful for the timely detection of signs of health and welfare
problems. Usually, this problem is addressed as a classification
problem, which typically requires manual annotation of behaviors. This
manual annotation can introduce noise into the data and may not always
be possible. This motivated us to address the problem as a time-series
forecasting problem in which the activity of an animal can be predicted.
In this work, different machine learning techniques were tested to
obtain activity patterns for Iberian pigs. In particular, we propose a
novel stacking ensemble learning approach that combines base learners
with meta-learners to obtain the final predictive model. Results confirm
the superior performance of the proposed method relative to the other
tested strategies. We also explored the possibility of using predictive
models trained on an animal to predict the activity of different animals
on the same farm. As expected, the predictive performance degrades in
this case, but it remains acceptable. The proposed method could be
integrated into a monitoring system that may have the potential to
transform the way farm animals are monitored, improving their health and
welfare conditions, for example, by allowing the early detection of a
possible health problem.
A. Gil-Gamboa and P. Paneque and O. Trull and A. Troncoso
Medium-term water consumption forecasting based on deep neural networks Journal Article
In: Expert Systems With Applications, vol. 247, pp. 123234, 2024.
Abstract | Links | BibTeX | Tags: deep learning, forecasting, time series
@article{ESWA2024,
title = {Medium-term water consumption forecasting based on deep neural networks},
author = {A. Gil-Gamboa and P. Paneque and O. Trull and A. Troncoso},
url = {https://www.sciencedirect.com/science/article/pii/S095741742400099X},
doi = {10.1016/j.eswa.2024.123234},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Expert Systems With Applications},
volume = {247},
pages = {123234},
abstract = {Water consumption forecasting is an essential tool for water management, as it allows for efficient planning and allocation of water resources, an undervalued but indispensable resource for all living beings. With the increasing demand for accurate and timely water forecasting, traditional forecasting methods are proving to be insufficient. Deep learning techniques, which have shown remarkable performance in a wide range of applications, offer a promising approach to address the challenges of water consumption forecasting. In this work, the use of deep learning models for medium-term water consumption forecasting of residential areas is explored. A deep feed-forward neural network is developed to predict water consumption of a company’s customers for the next quarter. First, customers are grouped according to their consumption as these customers include both household consumers and special consumers such as public swimming pools, sports halls or small industries. Then, a deep feed-forward neural network is designed for household customers by obtaining the optimal values for those hyperparameters that have a great influence on the network performance. Results are reported using a real-world dataset composed of the water consumption from 1999 to 2015 on a quarterly basis, corresponding to 3262 clients of a water supply company. Finally, the proposed algorithm is evaluated by comparing it with other reference algorithms including an LSTM network.},
keywords = {deep learning, forecasting, time series},
pubstate = {published},
tppubtype = {article}
}
Water consumption forecasting is an essential tool for water management, as it allows for efficient planning and allocation of water resources, an undervalued but indispensable resource for all living beings. With the increasing demand for accurate and timely water forecasting, traditional forecasting methods are proving to be insufficient. Deep learning techniques, which have shown remarkable performance in a wide range of applications, offer a promising approach to address the challenges of water consumption forecasting. In this work, the use of deep learning models for medium-term water consumption forecasting of residential areas is explored. A deep feed-forward neural network is developed to predict water consumption of a company’s customers for the next quarter. First, customers are grouped according to their consumption as these customers include both household consumers and special consumers such as public swimming pools, sports halls or small industries. Then, a deep feed-forward neural network is designed for household customers by obtaining the optimal values for those hyperparameters that have a great influence on the network performance. Results are reported using a real-world dataset composed of the water consumption from 1999 to 2015 on a quarterly basis, corresponding to 3262 clients of a water supply company. Finally, the proposed algorithm is evaluated by comparing it with other reference algorithms including an LSTM network.
J. F. Torres and M. Martinez-Ballesteros and A. Troncoso and F. Martínez-Álvarez
Special issue on Advances in Time Series Forecasting Journal Article
In: AIMS Mathematics, vol. 9, iss. 9, pp. 24163-24165, 2024.
@article{AIMS_Math2024,
title = {Special issue on Advances in Time Series Forecasting},
author = {J. F. Torres and M. Martinez-Ballesteros and A. Troncoso and F. Martínez-Álvarez},
url = {https://www.aimspress.com/article/doi/10.3934/math.20241174},
doi = {10.3934/math.20241174},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {AIMS Mathematics},
volume = {9},
issue = {9},
pages = {24163-24165},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Mártinez-Álvarez and A. Troncoso and G. Asencio-Cortés
From Simple to Complex: A Sequential Method for Enhancing Time Series Forecasting with Deep Learning Journal Article
In: Logic Journal of the IGPL, pp. jzae030, 2024.
Abstract | Links | BibTeX | Tags: deep learning, time series
@article{JIMENEZ-NAVARRO23a,
title = {From Simple to Complex: A Sequential Method for Enhancing Time Series Forecasting with Deep Learning},
author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Mártinez-Álvarez and A. Troncoso and G. Asencio-Cortés},
url = {https://academic.oup.com/jigpal/advance-article/doi/10.1093/jigpal/jzae030/7670726},
doi = {10.1093/jigpal/jzae030},
year = {2024},
date = {2024-01-01},
urldate = {2023-01-20},
journal = {Logic Journal of the IGPL},
pages = {jzae030},
abstract = {Time series forecasting is a well-known deep learning application field in which previous data are used to predict the future behavior of the series. Recently, several deep learning approaches have been proposed in which several nonlinear functions are applied to the input to obtain the output. In this paper, we introduce a novel method to improve the performance of deep learning models in time series forecasting. This method divides the model into hierarchies or levels from simpler to more complex ones. Simpler levels handle smoothed versions of the input, whereas the most complex level processes the original time series. This method follows the human learning process where general/simpler tasks are performed first, and afterward, more precise/harder ones are accomplished.Our proposed methodology has been applied to the LSTM architecture, showing remarkable performance in various time series. In addition, a comparison is reported including a standard LSTM and novel methods such as DeepAR, Temporal Fusion Transformer (TFT), NBEATS and Echo State Network (ESN).},
keywords = {deep learning, time series},
pubstate = {published},
tppubtype = {article}
}
Time series forecasting is a well-known deep learning application field in which previous data are used to predict the future behavior of the series. Recently, several deep learning approaches have been proposed in which several nonlinear functions are applied to the input to obtain the output. In this paper, we introduce a novel method to improve the performance of deep learning models in time series forecasting. This method divides the model into hierarchies or levels from simpler to more complex ones. Simpler levels handle smoothed versions of the input, whereas the most complex level processes the original time series. This method follows the human learning process where general/simpler tasks are performed first, and afterward, more precise/harder ones are accomplished.Our proposed methodology has been applied to the LSTM architecture, showing remarkable performance in various time series. In addition, a comparison is reported including a standard LSTM and novel methods such as DeepAR, Temporal Fusion Transformer (TFT), NBEATS and Echo State Network (ESN).
2023
A. R. Troncoso-García and I. S. Brito and A. Troncoso and F. Mártinez-Álvarez
Explainable hybrid deep learning and Coronavirus Optimization Algorithm for improving evapotranspiration forecasting Journal Article
In: Computers and Electronics in Agriculture, vol. 215, pp. 108387, 2023.
Abstract | Links | BibTeX | Tags: deep learning, forecasting, precision agriculture, XAI
@article{TRONCOSO-GARCIA23b,
title = {Explainable hybrid deep learning and Coronavirus Optimization Algorithm for improving evapotranspiration forecasting},
author = {A. R. Troncoso-García and I. S. Brito and A. Troncoso and F. Mártinez-Álvarez},
url = {https://www.sciencedirect.com/science/article/pii/S0168169923007755},
doi = {https://doi.org/10.1016/j.compag.2023.108387},
year = {2023},
date = {2023-11-08},
journal = {Computers and Electronics in Agriculture},
volume = {215},
pages = {108387},
abstract = {Reference evapotranspiration is a critical hydrological measurement closely associated with agriculture. Accurate forecasting is vital in effective water management and crop planning in sustainable agriculture. In this study, the future values of reference evapotranspiration are forecasted by applying a recurrent long shortterm memory neural network optimized using the Coronavirus Optimization Algorithm, a novel bioinspired metaheuristic based on the spread of COVID-19. The input data is sourced from the Sistema Agrometeorológico para a Gestão da Rega no Alentejo, in Portugal, with meteorological data such as air temperature or wind speed. Several baseline models are applied to the same problem to facilitate comparisons, including support vector machines, multi-layer perceptron, Lasso and decision tree. The results demonstrate the successful forecasting performance of the proposed model and its potential in this field. In turn, to gain deeper insights into the model’s inner workings, the SHapley Additive exPlanation tool is applied for explainability. Consequently, the study identifies the most relevant variables for reference evapotranspiration forecasting, including previously measured evapotranspiration values. Additionally, a univariable model is tested using historic evapotranspiration values as input, offering a comparable performance with a considerable reduction of computational time},
keywords = {deep learning, forecasting, precision agriculture, XAI},
pubstate = {published},
tppubtype = {article}
}
Reference evapotranspiration is a critical hydrological measurement closely associated with agriculture. Accurate forecasting is vital in effective water management and crop planning in sustainable agriculture. In this study, the future values of reference evapotranspiration are forecasted by applying a recurrent long shortterm memory neural network optimized using the Coronavirus Optimization Algorithm, a novel bioinspired metaheuristic based on the spread of COVID-19. The input data is sourced from the Sistema Agrometeorológico para a Gestão da Rega no Alentejo, in Portugal, with meteorological data such as air temperature or wind speed. Several baseline models are applied to the same problem to facilitate comparisons, including support vector machines, multi-layer perceptron, Lasso and decision tree. The results demonstrate the successful forecasting performance of the proposed model and its potential in this field. In turn, to gain deeper insights into the model’s inner workings, the SHapley Additive exPlanation tool is applied for explainability. Consequently, the study identifies the most relevant variables for reference evapotranspiration forecasting, including previously measured evapotranspiration values. Additionally, a univariable model is tested using historic evapotranspiration values as input, offering a comparable performance with a considerable reduction of computational time
D. Hadjout and A. Sebaa and J. F. Torres and F. Mártinez-Álvarez
Electricity consumption forecasting with outliers handling based on clustering and deep learning with application to the Algerian market Journal Article
In: Expert Systems with Applications, vol. 227, pp. 120123, 2023.
Abstract | Links | BibTeX | Tags: clustering, deep learning, energy, time series
@article{HADJOUT23,
title = {Electricity consumption forecasting with outliers handling based on clustering and deep learning with application to the Algerian market},
author = {D. Hadjout and A. Sebaa and J. F. Torres and F. Mártinez-Álvarez},
url = {https://www.sciencedirect.com/science/article/abs/pii/S0957417423006255},
doi = {https://doi.org/10.1016/j.eswa.2023.120123},
year = {2023},
date = {2023-10-01},
journal = {Expert Systems with Applications},
volume = {227},
pages = {120123},
abstract = {The reduction of electricity loss and the effective management of electricity demand are vital operations for production and distribution electricity enterprises. To achieve these goals, accurate forecasts of aggregate and individual electricity consumers are necessary. A novel multistep forecasting method is developed to forecast medium-term electricity consumption of the Algerian economic sector. The proposed method goes through the following three steps: cleaning steps, clustering steps and forecasting step of each cluster. The aim of the first step is to detect and then replace outliers. To complete the first phase, Robust Exponential and Holt-Winters Smoothing algorithms are adapted. Then, to carry out accurate forecasting at a lowest level, K-Shape and K-Means clustering methods are utilized to extract similarities and identify customer consumption patterns as a second step. The third step entails developing a deep learning model based on Gated Recurrent Units to forecast the electricity consumption in each cluster. To validate the proposed method, we compared our results to the most known methods in literature like Autoregressive Integrated Moving Average, Seasonal Grey Model, LSTM networks, Temporal Convolutional Networks and two ensemble models. The results of several experiments conducted with 2000 electricity consumers during 14 years from an Algeria province (Bejaia) demonstrate that the proposed method provides remarkable prediction performances. Thus, prediction performances of the K-Shape-based clustering method reach much higher prediction accuracy. According to the MAPE metric, the results of the best predictions are equal to 2.04%. It is also notable that 87% of the clients have a considerably low prediction error.},
keywords = {clustering, deep learning, energy, time series},
pubstate = {published},
tppubtype = {article}
}
The reduction of electricity loss and the effective management of electricity demand are vital operations for production and distribution electricity enterprises. To achieve these goals, accurate forecasts of aggregate and individual electricity consumers are necessary. A novel multistep forecasting method is developed to forecast medium-term electricity consumption of the Algerian economic sector. The proposed method goes through the following three steps: cleaning steps, clustering steps and forecasting step of each cluster. The aim of the first step is to detect and then replace outliers. To complete the first phase, Robust Exponential and Holt-Winters Smoothing algorithms are adapted. Then, to carry out accurate forecasting at a lowest level, K-Shape and K-Means clustering methods are utilized to extract similarities and identify customer consumption patterns as a second step. The third step entails developing a deep learning model based on Gated Recurrent Units to forecast the electricity consumption in each cluster. To validate the proposed method, we compared our results to the most known methods in literature like Autoregressive Integrated Moving Average, Seasonal Grey Model, LSTM networks, Temporal Convolutional Networks and two ensemble models. The results of several experiments conducted with 2000 electricity consumers during 14 years from an Algeria province (Bejaia) demonstrate that the proposed method provides remarkable prediction performances. Thus, prediction performances of the K-Shape-based clustering method reach much higher prediction accuracy. According to the MAPE metric, the results of the best predictions are equal to 2.04%. It is also notable that 87% of the clients have a considerably low prediction error.
J. F. Torres and S. Valencia and F. Martínez-Álvarez and N. Hoyos
Predicting Wildfires in the Caribbean Using Multi-source Satellite Data and Deep Learning Conference
IWANN 17th International Work-Conference on Artificial Neural Networks, vol. 14135, Lecture Notes in Computer Science 2023.
Links | BibTeX | Tags: deep learning, natural disasters, time series
@conference{TORRES23_IWANN,
title = {Predicting Wildfires in the Caribbean Using Multi-source Satellite Data and Deep Learning},
author = {J. F. Torres and S. Valencia and F. Martínez-Álvarez and N. Hoyos},
url = {https://link.springer.com/chapter/10.1007/978-3-031-43078-7_1},
doi = {https://doi.org/10.1007/978-3-031-43078-7_1},
year = {2023},
date = {2023-09-30},
booktitle = {IWANN 17th International Work-Conference on Artificial Neural Networks},
volume = {14135},
pages = {3-14},
series = {Lecture Notes in Computer Science},
keywords = {deep learning, natural disasters, time series},
pubstate = {published},
tppubtype = {conference}
}
M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Martínez-Álvarez and G. Asencio-Cortés
Embedded Temporal Feature Selection for Time Series Forecasting Using Deep Learning Conference
IWANN 17th International Work-Conference on Artificial Neural Networks, vol. 14135, Lecture Notes in Computer Science 2023.
Links | BibTeX | Tags: deep learning, feature selection, time series
@conference{JIMENEZ-NAVARRO23_IWANN,
title = {Embedded Temporal Feature Selection for Time Series Forecasting Using Deep Learning},
author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Martínez-Álvarez and G. Asencio-Cortés},
url = {https://link.springer.com/chapter/10.1007/978-3-031-43078-7_2},
doi = {https://doi.org/10.1007/978-3-031-43078-7_2},
year = {2023},
date = {2023-09-30},
booktitle = {IWANN 17th International Work-Conference on Artificial Neural Networks},
volume = {14135},
pages = {15-26},
series = {Lecture Notes in Computer Science},
keywords = {deep learning, feature selection, time series},
pubstate = {published},
tppubtype = {conference}
}
P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado
Springer, vol. 749, 2023, ISBN: 978-3-031-42529-5.
Links | BibTeX | Tags: big data, clustering, deep learning, IoT
@proceedings{SOCO2023a,
title = {Proceedings of the 18th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2023) Salamanca, Spain, September 5-7, 2023, volume 1},
author = {P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado},
editor = {P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado},
url = {https://link.springer.com/book/10.1007/978-3-031-42529-5},
doi = {https://doi.org/10.1007/978-3-031-42529-5},
isbn = {978-3-031-42529-5},
year = {2023},
date = {2023-09-05},
volume = {749},
publisher = {Springer},
series = {Lecture Notes in Networks and Systems},
keywords = {big data, clustering, deep learning, IoT},
pubstate = {published},
tppubtype = {proceedings}
}
P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado
Springer, vol. 750, 2023, ISBN: 978-3-031-42536-3.
Links | BibTeX | Tags: big data, clustering, deep learning, IoT
@proceedings{SOCO2023b,
title = {Proceedings of the 18th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2023) Salamanca, Spain, September 5-7, 2023, volume 2},
author = {P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado},
editor = {P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado},
url = {https://link.springer.com/book/10.1007/978-3-031-42536-3},
doi = {10.1007/978-3-030-20055-8},
isbn = {978-3-031-42536-3},
year = {2023},
date = {2023-09-05},
volume = {750},
publisher = {Springer},
series = {Lecture Notes in Networks and Systems},
keywords = {big data, clustering, deep learning, IoT},
pubstate = {published},
tppubtype = {proceedings}
}
P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado
Springer, vol. 14001, 2023, ISBN: 978-3-031-40725-3.
Links | BibTeX | Tags: big data, clustering, deep learning, IoT
@proceedings{HAIS2023,
title = {Proceedings of the 18th International Conference on Hybrid Artificial Intelligent Systems (HAIS 2023) Salamanca, Spain, September 5-7, 2023},
author = {P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado},
editor = {P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado},
url = {https://link.springer.com/book/10.1007/978-3-031-40725-3},
doi = {https://doi.org/10.1007/978-3-031-40725-3},
isbn = {978-3-031-40725-3},
year = {2023},
date = {2023-09-05},
volume = {14001},
publisher = {Springer},
series = {Lecture Notes in Artificial Intelligence},
keywords = {big data, clustering, deep learning, IoT},
pubstate = {published},
tppubtype = {proceedings}
}
P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado
Springer, vol. 748, 2023, ISBN: 978-3-031-42519-6.
Links | BibTeX | Tags: big data, clustering
@proceedings{CISIS-ICEUTE2023,
title = {Proceedings of the International Joint Conference 16th International Conference on Computational Intelligence in Security for Information Systems (CISIS 2023) 14th International Conference on EUropean Transnational Education (ICEUTE 2023). Salamanca, Spain, September 5-7, 2023},
author = {P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado},
editor = {P. García-Bringas and H. Pérez-García and F. J. Martínez de Pisón and F. Martínez-Álvarez and A. Troncoso and Á. Herrero and J. L. Calvo-Rolle and H. Quintián and E. Corchado},
url = {https://link.springer.com/book/10.1007/978-3-031-42519-6},
doi = {https://doi.org/10.1007/978-3-031-42519-6},
isbn = {978-3-031-42519-6},
year = {2023},
date = {2023-09-05},
volume = {748},
publisher = {Springer},
series = {Lecture Notes in Networks and Systems},
keywords = {big data, clustering},
pubstate = {published},
tppubtype = {proceedings}
}
A. Vellinger and J. F. Torres and F. Divina and W. Vanhoof
Neuroevolutionary Transfer Learning for Time Series Forecasting Conference
SOCO International Conference on Soft Computing Models in Industrial and Environmental Applications, vol. 749, Lecture Notes in Networks and Systems 2023.
Links | BibTeX | Tags: deep learning, forecasting, time series, transfer learning
@conference{VELLINGER23,
title = {Neuroevolutionary Transfer Learning for Time Series Forecasting},
author = {A. Vellinger and J. F. Torres and F. Divina and W. Vanhoof},
doi = {https://doi.org/10.1007/978-3-031-42529-5_21},
year = {2023},
date = {2023-08-31},
booktitle = {SOCO International Conference on Soft Computing Models in Industrial and Environmental Applications},
volume = {749},
pages = {219-228},
series = {Lecture Notes in Networks and Systems},
keywords = {deep learning, forecasting, time series, transfer learning},
pubstate = {published},
tppubtype = {conference}
}
M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Martínez-Álvarez and G. Asencio-Cortés
A New Deep Learning Architecture with Inductive Bias Balance for Oil Temperature Forecasting Journal Article
In: Journal of Big Data, vol. 10, pp. 80, 2023.
Abstract | Links | BibTeX | Tags: deep learning, time series
@article{JIMENEZ-NAVARRO23c,
title = {A New Deep Learning Architecture with Inductive Bias Balance for Oil Temperature Forecasting},
author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Martínez-Álvarez and G. Asencio-Cortés},
url = {https://journalofbigdata.springeropen.com/articles/10.1186/s40537-023-00745-0},
doi = {https://doi.org/10.1186/s40537-023-00745-0},
year = {2023},
date = {2023-05-28},
journal = {Journal of Big Data},
volume = {10},
pages = {80},
abstract = {Ensuring the optimal performance of power transformers is a laborious task in which the insulation system plays a vital role in decreasing their deterioration. The insulation system uses insulating oil to control temperature, as high temperatures can reduce the lifetime of the transformers and lead to expensive maintenance. Deep learning architectures have been demonstrated remarkable results in various fields. However, this improvement often comes at the cost of increased computing resources, which, in turn, increases the carbon footprint and hinders the optimization of architectures. In this study, we introduce a novel deep learning architecture that achieves a comparable efficacy to the best existing architectures in transformer oil temperature forecasting while improving efficiency. Effective forecasting can help prevent high temperatures and monitor the future condition of power transformers, thereby reducing unnecessary waste. To balance the inductive bias in our architecture, we propose the Smooth Residual Block, which divides the original problem into multiple subproblems to obtain different representations of the time series, collaboratively achieving the final forecasting. We applied our architecture to the Electricity Transformer datasets, which obtain transformer insulating oil temperature measures from two transformers in China. The results showed a 13% improvement in MSE and a 57% improvement in performance compared to the best current architectures, to the best of our knowledge. Moreover, we analyzed the architecture behavior to gain an intuitive understanding of the achieved solution.},
keywords = {deep learning, time series},
pubstate = {published},
tppubtype = {article}
}
Ensuring the optimal performance of power transformers is a laborious task in which the insulation system plays a vital role in decreasing their deterioration. The insulation system uses insulating oil to control temperature, as high temperatures can reduce the lifetime of the transformers and lead to expensive maintenance. Deep learning architectures have been demonstrated remarkable results in various fields. However, this improvement often comes at the cost of increased computing resources, which, in turn, increases the carbon footprint and hinders the optimization of architectures. In this study, we introduce a novel deep learning architecture that achieves a comparable efficacy to the best existing architectures in transformer oil temperature forecasting while improving efficiency. Effective forecasting can help prevent high temperatures and monitor the future condition of power transformers, thereby reducing unnecessary waste. To balance the inductive bias in our architecture, we propose the Smooth Residual Block, which divides the original problem into multiple subproblems to obtain different representations of the time series, collaboratively achieving the final forecasting. We applied our architecture to the Electricity Transformer datasets, which obtain transformer insulating oil temperature measures from two transformers in China. The results showed a 13% improvement in MSE and a 57% improvement in performance compared to the best current architectures, to the best of our knowledge. Moreover, we analyzed the architecture behavior to gain an intuitive understanding of the achieved solution.
M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Martínez-Álvarez and G. Asencio-Cortés
PHILNet: A Novel Efficient Approach for Time Series Forecasting using Deep Learning Journal Article
In: Information Sciences, vol. 632, pp. 815-832, 2023.
Abstract | Links | BibTeX | Tags: deep learning, time series
@article{JIMENEZ-NAVARRO23b,
title = {PHILNet: A Novel Efficient Approach for Time Series Forecasting using Deep Learning},
author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Martínez-Álvarez and G. Asencio-Cortés},
url = {https://doi.org/10.1016/j.ins.2023.03.021},
doi = {https://www.sciencedirect.com/science/article/pii/S0020025523003183?via%3Dihub},
year = {2023},
date = {2023-03-03},
journal = {Information Sciences},
volume = {632},
pages = {815-832},
abstract = {Time series is one of the most common data types in the industry nowadays. Forecasting the future of a time series behavior can be useful in planning ahead, saving time, resources, and helping avoid undesired scenarios. To make the forecasting, historical data is utilized due to the causal nature of the time series. Several deep learning algorithms have been presented in this area, where the input is processed through a series of non-linear functions to produce the output. We present a novel strategy to improve the performance of deep learning models in time series forecasting in terms of efficiency while reaching similar effectiveness. This approach separates the model into levels, starting with the easiest and continuing to the most difficult. The simpler levels deal with smoothed versions of the input, whereas the most sophisticated level deals with the raw data. This strategy seeks to mimic the human learning process, in which basic tasks are completed initially, followed by more precise and sophisticated ones. Our method achieved promising results, obtaining a 35% improvement in mean squared error and a 2.6 time decrease in training time compared with the best models found in a variety of time series.},
keywords = {deep learning, time series},
pubstate = {published},
tppubtype = {article}
}
Time series is one of the most common data types in the industry nowadays. Forecasting the future of a time series behavior can be useful in planning ahead, saving time, resources, and helping avoid undesired scenarios. To make the forecasting, historical data is utilized due to the causal nature of the time series. Several deep learning algorithms have been presented in this area, where the input is processed through a series of non-linear functions to produce the output. We present a novel strategy to improve the performance of deep learning models in time series forecasting in terms of efficiency while reaching similar effectiveness. This approach separates the model into levels, starting with the easiest and continuing to the most difficult. The simpler levels deal with smoothed versions of the input, whereas the most sophisticated level deals with the raw data. This strategy seeks to mimic the human learning process, in which basic tasks are completed initially, followed by more precise and sophisticated ones. Our method achieved promising results, obtaining a 35% improvement in mean squared error and a 2.6 time decrease in training time compared with the best models found in a variety of time series.
O. S. Mazari and A. Sebaa and J. L. Amaro-Mellado and F. Martínez-Álvarez
Creating a homogenized earthquake catalog for Algeria and mapping the main seismic parameters using a geographic information system Journal Article
In: Journal of African Earth Sciences, vol. 201, pp. 104859, 2023.
Abstract | Links | BibTeX | Tags: natural disasters
@article{MAZARI23,
title = {Creating a homogenized earthquake catalog for Algeria and mapping the main seismic parameters using a geographic information system},
author = {O. S. Mazari and A. Sebaa and J. L. Amaro-Mellado and F. Martínez-Álvarez},
url = {https://www.sciencedirect.com/science/article/pii/S1464343X23000687},
doi = {https://doi.org/10.1016/j.jafrearsci.2023.104895},
year = {2023},
date = {2023-03-03},
journal = {Journal of African Earth Sciences},
volume = {201},
pages = {104859},
abstract = {A homogeneous earthquake catalog is an essential instrument to study earthquake occurrence patterns, employing diverse engineering applications. In this paper, we describe a series of compilation and processing steps to compile an updated earthquake catalog for Algeria, a North African country with relatively high seismic activity. The procedure consisted of several steps. First, a range of reliable catalogs were considered; second, the data was integrated and refined; third, magnitudes are homogenized from different kinds of magnitudes into moment magnitude (M_w); declustering is then performed; and, finally, the magnitude-year completeness was estimated. The resulting Algeria catalog is bounded by the geographical limits (19° - 38.5° N and 9.5° W - 12.5° E), and covers the 1960-2020 period. It includes 4021 seismic events, reported up to M_w 7.1. We also calculate a set of seismic parameters, namely M_max and b-value, and mapped them using a geographic information system. Thus, the territory is divided into cells based on different grids to conduct the analysis. The results of the seismic parameters mapping are discussed, highlighting significant details. Several cells presented a M_max between 6.0 and 7.1. Regarding the b-value, two regions (Oran and Constantine) presented a high b-value, implying low-stress areas, and three regions (Algiers, Batna, and Chlef) a low b-value (0.65- 0.85), suggesting high-stress areas. Finally, we suggest some recommendations for future seismic hazard assessment studies.},
keywords = {natural disasters},
pubstate = {published},
tppubtype = {article}
}
A homogeneous earthquake catalog is an essential instrument to study earthquake occurrence patterns, employing diverse engineering applications. In this paper, we describe a series of compilation and processing steps to compile an updated earthquake catalog for Algeria, a North African country with relatively high seismic activity. The procedure consisted of several steps. First, a range of reliable catalogs were considered; second, the data was integrated and refined; third, magnitudes are homogenized from different kinds of magnitudes into moment magnitude (M_w); declustering is then performed; and, finally, the magnitude-year completeness was estimated. The resulting Algeria catalog is bounded by the geographical limits (19° - 38.5° N and 9.5° W - 12.5° E), and covers the 1960-2020 period. It includes 4021 seismic events, reported up to M_w 7.1. We also calculate a set of seismic parameters, namely M_max and b-value, and mapped them using a geographic information system. Thus, the territory is divided into cells based on different grids to conduct the analysis. The results of the seismic parameters mapping are discussed, highlighting significant details. Several cells presented a M_max between 6.0 and 7.1. Regarding the b-value, two regions (Oran and Constantine) presented a high b-value, implying low-stress areas, and three regions (Algiers, Batna, and Chlef) a low b-value (0.65- 0.85), suggesting high-stress areas. Finally, we suggest some recommendations for future seismic hazard assessment studies.
D. Azzouguer and A. Sebaa and D. Hadjout and F. Martínez-Álvarez
IEEE International Conference on Advanced Systems and Emergent Technologies, 2023.
Abstract | Links | BibTeX | Tags: energy, time series
@conference{AZZOUGUER23,
title = {Fraud Detection of Electricity Consumption using Robust Exponential and Holt-Winters Smoothing method},
author = {D. Azzouguer and A. Sebaa and D. Hadjout and F. Martínez-Álvarez},
url = {https://ieeexplore.ieee.org/document/10150645},
doi = {10.1109/IC_ASET58101.2023.10150645},
year = {2023},
date = {2023-02-20},
booktitle = {IEEE International Conference on Advanced Systems and Emergent Technologies},
abstract = {Non-technical losses (NTL), especially fraud detection is very important for electricity distribution enterprises. Fraud detection allows for maximizing the effective economic return for such enterprises. This paper provides an electricity fraud detection approach based on robust exponential and Holt-Winters Smoothing methods. The proposed approach is a procedure that aims to discover the fraudulent behavior of electricity consumers and goes through three crucial steps: (1) the prediction of monthly consumption, (2) the detection of abnormal consumption of electrical meters, and (3) the detection of fraud cases of economic customers. The proposed model was trained and evaluated. Its experimental validation is achieved by using a large dataset of real users from the Algerian economic sector with almost 2000 clients and 14 years of monthly electricity consumption. The proposed solution revealed good performance compared to the literature and the comparison with the models implemented in this article: SARIMA for prediction and two sigma for anomaly detection. The results show highly efficient and realistic countermeasures to fraud detection, which leads us to say that this method is robust and can enhance company profit.},
keywords = {energy, time series},
pubstate = {published},
tppubtype = {conference}
}
Non-technical losses (NTL), especially fraud detection is very important for electricity distribution enterprises. Fraud detection allows for maximizing the effective economic return for such enterprises. This paper provides an electricity fraud detection approach based on robust exponential and Holt-Winters Smoothing methods. The proposed approach is a procedure that aims to discover the fraudulent behavior of electricity consumers and goes through three crucial steps: (1) the prediction of monthly consumption, (2) the detection of abnormal consumption of electrical meters, and (3) the detection of fraud cases of economic customers. The proposed model was trained and evaluated. Its experimental validation is achieved by using a large dataset of real users from the Algerian economic sector with almost 2000 clients and 14 years of monthly electricity consumption. The proposed solution revealed good performance compared to the literature and the comparison with the models implemented in this article: SARIMA for prediction and two sigma for anomaly detection. The results show highly efficient and realistic countermeasures to fraud detection, which leads us to say that this method is robust and can enhance company profit.
E. T. Habtemariam and K. Kekeba and M. Martínez-Ballesteros and F. Mártinez-Álvarez
A Bayesian Optimization-Based LSTM Model for Wind Power Forecasting in the Adama District, Ethiopia Journal Article
In: Energies, vol. 16, pp. 2317, 2023.
Abstract | Links | BibTeX | Tags: deep learning, time series
@article{EJIGU23,
title = {A Bayesian Optimization-Based LSTM Model for Wind Power Forecasting in the Adama District, Ethiopia},
author = {E. T. Habtemariam and K. Kekeba and M. Martínez-Ballesteros and F. Mártinez-Álvarez},
url = {https://www.mdpi.com/1996-1073/16/5/2317},
doi = {https://doi.org/10.3390/en16052317},
year = {2023},
date = {2023-02-19},
journal = {Energies},
volume = {16},
pages = {2317},
abstract = {Renewable energies such as solar and wind power have become promising sources of energy to address the increase in greenhouse gases caused by the use of fossil fuels and to resolve current energy crises. Integrating wind energy into a large-scale electric grid presents a significant challenge due to the high intermittency and nonlinear behavior of wind power. Accurate wind power forecasting is essential for safe and efficient integration into the grid system. Many prediction models have been developed to predict the uncertain and nonlinear time series of wind power, but most neglect the use of Bayesian optimization to optimize the hyperparameters while training deep learning algorithms. The efficiency of grid search strategies decreases as the number of hyperparameters increases, and computation time complexity becomes an issue. This paper presents a robust and optimized Long-Short Term Memory network for forecasting wind power generation in the day ahead in the context of Ethiopia's renewable energy sector. The proposal uses Bayesian optimization to find the best hyperparameter combination in a reasonable computation time. The results indicate that tuning hyperparameters using this metaheuristic prior to building deep learning models significantly improves the predictive performance of the models. The proposed models were evaluated using MAE, RMSE, and MAPE metrics and outperformed both the baseline models and the optimized Gated Recurrent Unit architecture.},
keywords = {deep learning, time series},
pubstate = {published},
tppubtype = {article}
}
Renewable energies such as solar and wind power have become promising sources of energy to address the increase in greenhouse gases caused by the use of fossil fuels and to resolve current energy crises. Integrating wind energy into a large-scale electric grid presents a significant challenge due to the high intermittency and nonlinear behavior of wind power. Accurate wind power forecasting is essential for safe and efficient integration into the grid system. Many prediction models have been developed to predict the uncertain and nonlinear time series of wind power, but most neglect the use of Bayesian optimization to optimize the hyperparameters while training deep learning algorithms. The efficiency of grid search strategies decreases as the number of hyperparameters increases, and computation time complexity becomes an issue. This paper presents a robust and optimized Long-Short Term Memory network for forecasting wind power generation in the day ahead in the context of Ethiopia's renewable energy sector. The proposal uses Bayesian optimization to find the best hyperparameter combination in a reasonable computation time. The results indicate that tuning hyperparameters using this metaheuristic prior to building deep learning models significantly improves the predictive performance of the models. The proposed models were evaluated using MAE, RMSE, and MAPE metrics and outperformed both the baseline models and the optimized Gated Recurrent Unit architecture.
A. M. Fernández and D. Gutiérrez-Avilés and A. Troncoso and F. Martínez-Álvarez
A new Apache Spark-based framework for big data streaming forecasting in IoT networks Journal Article
In: Journal of Supercomputing, vol. 79, pp. 11078–11100, 2023.
Abstract | Links | BibTeX | Tags: big data, IoT
@article{FERNANDEZ23,
title = {A new Apache Spark-based framework for big data streaming forecasting in IoT networks},
author = {A. M. Fernández and D. Gutiérrez-Avilés and A. Troncoso and F. Martínez-Álvarez},
url = {https://link.springer.com/article/10.1007/s11227-023-05100-x},
doi = {https://doi.org/10.1007/s11227-023-05100-x},
year = {2023},
date = {2023-02-02},
journal = {Journal of Supercomputing},
volume = {79},
pages = {11078–11100},
abstract = {Analyzing time-dependent data acquired in a continuous flow is a major challenge for various fields, such as big data and machine learning. Being able to analyze a large volume of data from various sources, such as sensors, networks, and the internet, is essential for improving the efficiency of our society's production processes. Additionally, this vast amount of data is collected dynamically in a continuous stream. The goal of this research is to provide a comprehensive framework for forecasting big data streams from Internet of Things networks and serve as a guide for designing and deploying other third-party solutions. Hence, a new framework for time series forecasting in a big data streaming scenario, using data collected from Internet of Things networks, is presented. This framework comprises of five main modules: Internet of Things network design and deployment, big data streaming architecture, stream data modeling method, big data forecasting method, and a comprehensive real-world application scenario, consisting of a physical Internet of Things network feeding the big data streaming architecture, being the linear regression the algorithm used for illustrative purposes. Comparison with other frameworks reveals that this is the first framework that incorporates and integrates all the aforementioned modules.},
keywords = {big data, IoT},
pubstate = {published},
tppubtype = {article}
}
Analyzing time-dependent data acquired in a continuous flow is a major challenge for various fields, such as big data and machine learning. Being able to analyze a large volume of data from various sources, such as sensors, networks, and the internet, is essential for improving the efficiency of our society's production processes. Additionally, this vast amount of data is collected dynamically in a continuous stream. The goal of this research is to provide a comprehensive framework for forecasting big data streams from Internet of Things networks and serve as a guide for designing and deploying other third-party solutions. Hence, a new framework for time series forecasting in a big data streaming scenario, using data collected from Internet of Things networks, is presented. This framework comprises of five main modules: Internet of Things network design and deployment, big data streaming architecture, stream data modeling method, big data forecasting method, and a comprehensive real-world application scenario, consisting of a physical Internet of Things network feeding the big data streaming architecture, being the linear regression the algorithm used for illustrative purposes. Comparison with other frameworks reveals that this is the first framework that incorporates and integrates all the aforementioned modules.
A. R. Troncoso-García and M. Martínez-Ballesteros and F. Mártinez-Álvarez and A. Troncoso
A new approach based on association rules to add explainability to time series forecasting models Journal Article
In: Information Fusion, vol. 94, pp. 169-180, 2023.
Abstract | Links | BibTeX | Tags: association rules, time series, XAI
@article{TRONCOSO-GARCIA23,
title = {A new approach based on association rules to add explainability to time series forecasting models},
author = {A. R. Troncoso-García and M. Martínez-Ballesteros and F. Mártinez-Álvarez and A. Troncoso},
url = {https://www.sciencedirect.com/science/article/pii/S1566253523000295},
doi = {10.1016/j.inffus.2023.01.021},
year = {2023},
date = {2023-01-22},
journal = {Information Fusion},
volume = {94},
pages = {169-180},
abstract = {Machine learning and deep learning have become the most useful and powerful tools in the last years to mine information from large datasets. Despite the successful application to many research fields, it is widely known that some of these solutions based on artificial intelligence are considered black-box models, meaning that most experts find difficult to explain and interpret the models and why they generate such outputs. In this context, explainable artificial intelligence is emerging with the aim of providing black-box models with sufficient interpretability. Thus, models could be easily understood and further applied. This work proposes a novel method to explain black-box models, by using numeric association rules to explain and interpret multi-step time series forecasting models. Thus, a multi-objective algorithm is used to discover quantitative association rules from the target model. Then, visual explanation techniques are applied to make the rules more interpretable. Data from Spanish electricity energy consumption has been used to assess the suitability of the proposal.},
keywords = {association rules, time series, XAI},
pubstate = {published},
tppubtype = {article}
}
Machine learning and deep learning have become the most useful and powerful tools in the last years to mine information from large datasets. Despite the successful application to many research fields, it is widely known that some of these solutions based on artificial intelligence are considered black-box models, meaning that most experts find difficult to explain and interpret the models and why they generate such outputs. In this context, explainable artificial intelligence is emerging with the aim of providing black-box models with sufficient interpretability. Thus, models could be easily understood and further applied. This work proposes a novel method to explain black-box models, by using numeric association rules to explain and interpret multi-step time series forecasting models. Thus, a multi-objective algorithm is used to discover quantitative association rules from the target model. Then, visual explanation techniques are applied to make the rules more interpretable. Data from Spanish electricity energy consumption has been used to assess the suitability of the proposal.
M. J. Jiménez-Navarro and M. Martínez-Ballesteros and I. S. Brito and F. Martínez-Álvarez and G. Asencio-Cortés
SAC 38th Annual ACM Symposium on Applied Computing, 2023.
Abstract | Links | BibTeX | Tags: deep learning, precision agriculture, time series
@conference{EVAPOCVOA23,
title = {A bioinspired ensemble approach for multi-horizon reference evapotranspiration forecasting in Portugal},
author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and I. S. Brito and F. Martínez-Álvarez and G. Asencio-Cortés},
url = {https://dl.acm.org/doi/abs/10.1145/3555776.3578634},
doi = {https://doi.org/10.1145/3555776.3578634},
year = {2023},
date = {2023-01-01},
booktitle = {SAC 38th Annual ACM Symposium on Applied Computing},
pages = {441-448},
abstract = {The year 2022 was the driest year in Portugal since 1931 with 97% of territory in severe drought. Water is especially important for the agricultural sector in Portugal, as it represents 78% total consumption according to the Water Footprint report published in 2010. Reference evapotranspiration is essential due to its importance in optimal irrigation planning that reduces water consumption. This study analyzes and proposes a framework to forecast daily reference evapotranspiration at eight stations in Portugal from 2012 to 2022 without relying on public meteorological forecasts. The data include meteorological data obtained from sensors included in the stations. The goal is to perform a multi-horizon forecasting of reference evapotranspiration using the multiple related covariates. The framework combines the data processing and the analysis of several state-of-the-art forecasting methods including classical, linear, tree-based, artificial neural network and ensembles. Then, an ensemble of all trained models is proposed using a recent bioinspired metaheuristic named Coronavirus Optimization Algorithm to weight the predictions. The results in terms of MAE and MSE are reported, indicating that our approach achieved a MAE of 0.658.},
keywords = {deep learning, precision agriculture, time series},
pubstate = {published},
tppubtype = {conference}
}
The year 2022 was the driest year in Portugal since 1931 with 97% of territory in severe drought. Water is especially important for the agricultural sector in Portugal, as it represents 78% total consumption according to the Water Footprint report published in 2010. Reference evapotranspiration is essential due to its importance in optimal irrigation planning that reduces water consumption. This study analyzes and proposes a framework to forecast daily reference evapotranspiration at eight stations in Portugal from 2012 to 2022 without relying on public meteorological forecasts. The data include meteorological data obtained from sensors included in the stations. The goal is to perform a multi-horizon forecasting of reference evapotranspiration using the multiple related covariates. The framework combines the data processing and the analysis of several state-of-the-art forecasting methods including classical, linear, tree-based, artificial neural network and ensembles. Then, an ensemble of all trained models is proposed using a recent bioinspired metaheuristic named Coronavirus Optimization Algorithm to weight the predictions. The results in terms of MAE and MSE are reported, indicating that our approach achieved a MAE of 0.658.
L. Melgar-García, M. Hosseini and A. Troncoso
Identification of anomalies in urban sound data with Autoencoders Conference
HAIS 18th International Conference on Hybrid Artificial Intelligence Systems, Lecture Notes in Computer Science 2023.
BibTeX | Tags: deep learning, IoT, time series
@conference{HAIS23_Laura,
title = {Identification of anomalies in urban sound data with Autoencoders},
author = {L. Melgar-García, M. Hosseini and A. Troncoso},
year = {2023},
date = {2023-01-01},
booktitle = {HAIS 18th International Conference on Hybrid Artificial Intelligence Systems},
series = {Lecture Notes in Computer Science},
keywords = {deep learning, IoT, time series},
pubstate = {published},
tppubtype = {conference}
}
E. Tefera and A. Troncoso and M. Martínez Ballesteros and F. Martínez-Álvarez
A New Hybrid CNN-LSTM for Wind Power Forecasting in Ethiopia Conference
HAIS 18th International Conference on Hybrid Artificial Intelligence Systems, Lecture Notes in Computer Science 2023.
BibTeX | Tags: deep learning, energy, time series
@conference{HAIS23_Ejigu,
title = {A New Hybrid CNN-LSTM for Wind Power Forecasting in Ethiopia},
author = {E. Tefera and A. Troncoso and M. Martínez Ballesteros and F. Martínez-Álvarez},
year = {2023},
date = {2023-01-01},
booktitle = {HAIS 18th International Conference on Hybrid Artificial Intelligence Systems},
series = {Lecture Notes in Computer Science},
keywords = {deep learning, energy, time series},
pubstate = {published},
tppubtype = {conference}
}
L. Melgar-García, Á. Troncoso-García, D. Gutiérrez-Avilés, J. F. Torres and A. Troncoso
Explainable Artificial Intelligence for Education: A Real Case of a University Subject Switched to Python Conference
ICEUTE 14th International Conference on European Transnational Educational, Lecture Notes in Networks and Systems 2023.
@conference{ICEUTE23_Laura,
title = {Explainable Artificial Intelligence for Education: A Real Case of a University Subject Switched to Python},
author = {L. Melgar-García, Á. Troncoso-García, D. Gutiérrez-Avilés, J. F. Torres and A. Troncoso},
year = {2023},
date = {2023-01-01},
booktitle = {ICEUTE 14th International Conference on European Transnational Educational},
series = {Lecture Notes in Networks and Systems},
keywords = {education, XAI},
pubstate = {published},
tppubtype = {conference}
}
A. M. Chacón-Maldonado and A.R. Troncoso-García and F. Martínez-Álvarez, G. Asencio-Cortés and A. Troncoso
Olive oil fly population pest forecasting using explainable deep learning Conference
SOCO 18th International Conference on Soft Computing Models in Industrial and Environmental Applications , Lecture Notes in Networks and Systems 2023.
BibTeX | Tags: precision agriculture, XAI
@conference{SOCO23_Andres,
title = {Olive oil fly population pest forecasting using explainable deep learning},
author = {A. M. Chacón-Maldonado and A.R. Troncoso-García and F. Martínez-Álvarez, G. Asencio-Cortés and A. Troncoso},
year = {2023},
date = {2023-01-01},
booktitle = {SOCO 18th International Conference on Soft Computing Models in Industrial and Environmental Applications },
series = {Lecture Notes in Networks and Systems},
keywords = {precision agriculture, XAI},
pubstate = {published},
tppubtype = {conference}
}
P. Casas-Gómez and F. Martínez-Álvarez and A. Troncoso and J. C. Linares-Calderón
Machine Learning Approaches for Predicting Tree Growth Trends based on Basal Area Increment Conference
SOCO 18th International Conference on Soft Computing Models in Industrial and Environmental Applications, Lecture Notes in Networks and Systems 2023.
BibTeX | Tags: time series
@conference{SOCO22_Pablo,
title = {Machine Learning Approaches for Predicting Tree Growth Trends based on Basal Area Increment},
author = {P. Casas-Gómez and F. Martínez-Álvarez and A. Troncoso and J. C. Linares-Calderón},
year = {2023},
date = {2023-01-01},
booktitle = {SOCO 18th International Conference on Soft Computing Models in Industrial and Environmental Applications},
series = {Lecture Notes in Networks and Systems},
keywords = {time series},
pubstate = {published},
tppubtype = {conference}
}
L. Melgar-García and F. Martínez-Álvarez and D. T. Bui and A. Troncoso
In: International Journal of Digital Earth, vol. 16, no. 1, pp. 3661-3679, 2023.
Links | BibTeX | Tags: deep learning, natural disasters
@article{Melgar2023c,
title = {A novel semantic segmentation approach based on U-Net, WU-Net, and U-Net++ deep learning for predicting areas sensitive to pluvial flood at tropical area},
author = {L. Melgar-García and F. Martínez-Álvarez and D. T. Bui and A. Troncoso},
url = {https://www.tandfonline.com/doi/full/10.1080/17538947.2023.2252401},
doi = {https://doi.org/10.1080/17538947.2023.2252401},
year = {2023},
date = {2023-01-01},
journal = {International Journal of Digital Earth},
volume = {16},
number = {1},
pages = {3661-3679},
keywords = {deep learning, natural disasters},
pubstate = {published},
tppubtype = {article}
}
M. Vázquez-Marrufo and E. Sarrias-Arrabal and M. García-Torres and R. Martín-Clemente and G. Izquierdo
A systematic review of the application of machine-learning algorithms in multiple sclerosis Journal Article
In: Neurología (English Edition), 2023.
Abstract | Links | BibTeX | Tags: bioinformatics
@article{vazquez2022systematic,
title = {A systematic review of the application of machine-learning algorithms in multiple sclerosis},
author = {M. Vázquez-Marrufo and E. Sarrias-Arrabal and M. García-Torres and R. Martín-Clemente and G. Izquierdo},
url = {https://www.sciencedirect.com/science/article/pii/S217358082200075X},
doi = {10.1016/j.nrleng.2020.10.013},
year = {2023},
date = {2023-01-01},
journal = {Neurología (English Edition)},
publisher = {Elsevier},
abstract = {Introduction: The applications of artificial intelligence, and in particular automatic learning or “machine learning” (ML), constitute both a challenge and a great opportunity in numerous scientific, technical, and clinical disciplines. Specific applications in the study of multiple sclerosis (MS) have been no exception, and constitute an area of increasing interest in recent years. Objective: We present a systematic review of the application of ML algorithms in MS. Materials and methods: We used the PubMed search engine, which allows free access to the MEDLINE medical database, to identify studies including the keywords “machine learning” and “multiple sclerosis.” We excluded review articles, studies written in languages other than English or Spanish, and studies that were mainly technical and did not specifically apply to MS. The final selection included 76 articles, and 38 were rejected. Conclusions: After the review process, we established 4 main applications of ML in MS: 1) classifying MS subtypes; 2) distinguishing patients with MS from healthy controls and individuals with other diseases; 3) predicting progression and response to therapeutic interventions; and 4) other applications. Results found to date have shown that ML algorithms may offer great support for health professionals both in clinical settings and in research into MS.},
keywords = {bioinformatics},
pubstate = {published},
tppubtype = {article}
}
Introduction: The applications of artificial intelligence, and in particular automatic learning or “machine learning” (ML), constitute both a challenge and a great opportunity in numerous scientific, technical, and clinical disciplines. Specific applications in the study of multiple sclerosis (MS) have been no exception, and constitute an area of increasing interest in recent years. Objective: We present a systematic review of the application of ML algorithms in MS. Materials and methods: We used the PubMed search engine, which allows free access to the MEDLINE medical database, to identify studies including the keywords “machine learning” and “multiple sclerosis.” We excluded review articles, studies written in languages other than English or Spanish, and studies that were mainly technical and did not specifically apply to MS. The final selection included 76 articles, and 38 were rejected. Conclusions: After the review process, we established 4 main applications of ML in MS: 1) classifying MS subtypes; 2) distinguishing patients with MS from healthy controls and individuals with other diseases; 3) predicting progression and response to therapeutic interventions; and 4) other applications. Results found to date have shown that ML algorithms may offer great support for health professionals both in clinical settings and in research into MS.
G. Sosa-Cabrera and S. Gómez-Guerrero and M. García-Torres and C. E. Schaerer
Feature selection: A perspective on inter-attribute cooperation Journal Article
In: International Journal of Data Science and Analytics, pp. 1–13, 2023.
Abstract | Links | BibTeX | Tags: feature selection
@article{sosa2023feature,
title = {Feature selection: A perspective on inter-attribute cooperation},
author = {G. Sosa-Cabrera and S. Gómez-Guerrero and M. García-Torres and C. E. Schaerer},
url = {https://link.springer.com/article/10.1007/s41060-023-00439-z},
doi = {10.1007/s41060-023-00439-z},
year = {2023},
date = {2023-01-01},
journal = {International Journal of Data Science and Analytics},
pages = {1--13},
publisher = {Springer},
abstract = {High-dimensional datasets depict a challenge for learning tasks in data mining and machine learning. Feature selection is an effective technique in dealing with dimensionality reduction. It is often an essential data processing step prior to applying a learning algorithm. Over the decades, filter feature selection methods have evolved from simple univariate relevance ranking algorithms to more sophisticated relevance-redundancy trade-offs and to multivariate dependencies-based approaches in recent years. This tendency to capture multivariate dependence aims at obtaining unique information about the class from the intercooperation among features. This paper presents a comprehensive survey of the state-of-the-art work on filter feature selection methods assisted by feature intercooperation, and summarizes the contributions of different approaches found in the literature. Furthermore, current issues and challenges are introduced to identify promising future research and development.},
keywords = {feature selection},
pubstate = {published},
tppubtype = {article}
}
High-dimensional datasets depict a challenge for learning tasks in data mining and machine learning. Feature selection is an effective technique in dealing with dimensionality reduction. It is often an essential data processing step prior to applying a learning algorithm. Over the decades, filter feature selection methods have evolved from simple univariate relevance ranking algorithms to more sophisticated relevance-redundancy trade-offs and to multivariate dependencies-based approaches in recent years. This tendency to capture multivariate dependence aims at obtaining unique information about the class from the intercooperation among features. This paper presents a comprehensive survey of the state-of-the-art work on filter feature selection methods assisted by feature intercooperation, and summarizes the contributions of different approaches found in the literature. Furthermore, current issues and challenges are introduced to identify promising future research and development.
P. Jiménez-Herrera and L. Melgar-García and G. Asencio-Cortés and A. Troncoso
Streaming big time series forecasting based on nearest similar patterns with application to energy consumption Journal Article
In: Logic Journal of the IGPL, vol. 31, no. 2, pp. 255-270, 2023.
Abstract | Links | BibTeX | Tags: energy, IoT, time series
@article{jimenez2023,
title = {Streaming big time series forecasting based on nearest similar patterns with application to energy consumption},
author = {P. Jiménez-Herrera and L. Melgar-García and G. Asencio-Cortés and A. Troncoso},
url = {https://academic.oup.com/jigpal/advance-article-abstract/doi/10.1093/jigpal/jzac017/6534493?redirectedFrom=fulltext},
doi = {https://doi.org/10.1093/jigpal/jzac017},
year = {2023},
date = {2023-01-01},
journal = {Logic Journal of the IGPL},
volume = {31},
number = {2},
pages = {255-270},
abstract = {This work presents a novel approach to forecast streaming big time series based on nearest similar patterns. This approach combines a clustering algorithm with a classifier and the nearest neighbors algorithm. It presents two separate stages: offline and online. The offline phase is for training and finding the best models for clustering, classification and the nearest neighbors algorithm. The online phase is to predict big time series in real time. In the offline phase, data are divided into clusters and a forecasting model based on the nearest neighbors is trained for each cluster. In addition, a classifier is trained using the cluster assignments previously generated by the clustering algorithm. In the online phase, the classifier predicts the cluster label of an instance, and the proper nearest neighbors model according to the predicted cluster label is applied to obtain the final prediction using the similar patterns. The algorithm is able to be updated incrementally for online learning from data streams. Results are reported using electricity consumption with a granularity of 10 minutes for 4-hour-ahead forecasting and compared with well-known online benchmark learners, showing a remarkable improvement in prediction accuracy.},
keywords = {energy, IoT, time series},
pubstate = {published},
tppubtype = {article}
}
This work presents a novel approach to forecast streaming big time series based on nearest similar patterns. This approach combines a clustering algorithm with a classifier and the nearest neighbors algorithm. It presents two separate stages: offline and online. The offline phase is for training and finding the best models for clustering, classification and the nearest neighbors algorithm. The online phase is to predict big time series in real time. In the offline phase, data are divided into clusters and a forecasting model based on the nearest neighbors is trained for each cluster. In addition, a classifier is trained using the cluster assignments previously generated by the clustering algorithm. In the online phase, the classifier predicts the cluster label of an instance, and the proper nearest neighbors model according to the predicted cluster label is applied to obtain the final prediction using the similar patterns. The algorithm is able to be updated incrementally for online learning from data streams. Results are reported using electricity consumption with a granularity of 10 minutes for 4-hour-ahead forecasting and compared with well-known online benchmark learners, showing a remarkable improvement in prediction accuracy.
O. Cardozo and V. Ojeda and R. Parra and J. C. Mello-Román and J. L. Noguera Vázquez and M. García-Torres and F. Divina and S. Grillo and C. Villalba and J. Facon
Dataset of fundus images for the diagnosis of ocular toxoplasmosis Journal Article
In: Data in Brief, pp. 109056, 2023.
Abstract | Links | BibTeX | Tags: bioinformatics
@article{cardozo2023dataset,
title = {Dataset of fundus images for the diagnosis of ocular toxoplasmosis},
author = {O. Cardozo and V. Ojeda and R. Parra and J. C. Mello-Román and J. L. Noguera Vázquez and M. García-Torres and F. Divina and S. Grillo and C. Villalba and J. Facon},
url = {https://www.sciencedirect.com/science/article/pii/S2352340923001749},
doi = {10.1016/j.dib.2023.109056},
year = {2023},
date = {2023-01-01},
journal = {Data in Brief},
pages = {109056},
publisher = {Elsevier},
abstract = {Toxoplasmosis chorioretinitis is commonly diagnosed by an ophthalmologist through the evaluation of the fundus images of a patient. Early detection of these lesions may help to prevent blindness. In this article we present a data set of fundus images labeled into three categories: healthy eye, inactive and active chorioretinitis. The dataset was developed by three ophthalmologists with expertise in toxoplasmosis detection using fundus images. The dataset will be of great use to researchers working on ophthalmic image analysis using artificial intelligence techniques for the automatic detection of toxoplasmosis chorioretinitis.},
keywords = {bioinformatics},
pubstate = {published},
tppubtype = {article}
}
Toxoplasmosis chorioretinitis is commonly diagnosed by an ophthalmologist through the evaluation of the fundus images of a patient. Early detection of these lesions may help to prevent blindness. In this article we present a data set of fundus images labeled into three categories: healthy eye, inactive and active chorioretinitis. The dataset was developed by three ophthalmologists with expertise in toxoplasmosis detection using fundus images. The dataset will be of great use to researchers working on ophthalmic image analysis using artificial intelligence techniques for the automatic detection of toxoplasmosis chorioretinitis.
M. García-Torres and R. Ruiz and F. Divina
Evolutionary feature selection on high dimensional data using a search space reduction approach Journal Article
In: Engineering Applications of Artificial Intelligence, vol. 117, pp. 105556, 2023.
Abstract | Links | BibTeX | Tags: big data, feature selection
@article{garcia2023evolutionary,
title = {Evolutionary feature selection on high dimensional data using a search space reduction approach},
author = {M. García-Torres and R. Ruiz and F. Divina},
url = {https://www.sciencedirect.com/science/article/pii/S0952197622005462},
doi = {10.1016/j.engappai.2022.105556},
year = {2023},
date = {2023-01-01},
journal = {Engineering Applications of Artificial Intelligence},
volume = {117},
pages = {105556},
publisher = {Elsevier},
abstract = {Feature selection is becoming more and more a challenging task due to the increase of the dimensionality of the data. The complexity of the interactions among features and the size of the search space make it unfeasible to find the optimal subset of features. In order to reduce the search space, feature grouping has arisen as an approach that allows to cluster feature according to the shared information about the class. On the other hand, metaheuristic algorithms have proven to achieve sub-optimal solutions within a reasonable time. In this work we propose a Scatter Search (SS) strategy that uses feature grouping to generate an initial population comprised of diverse and high quality solutions. Solutions are then evolved by applying random mechanisms in combination with the feature group structure, with the objective of maintaining during the search a population of good and, at the same time, as diverse as possible solutions. Not only does the proposed strategy provide the best subset of features found but it also reduces the redundancy structure of the data. We test the strategy on high dimensional data from biomedical and text-mining domains. The results are compared with those obtained by other adaptations of SS and other popular strategies. Results show that the proposed strategy can find, on average, the smallest subsets of features without degrading the performance of the classifier.},
keywords = {big data, feature selection},
pubstate = {published},
tppubtype = {article}
}
Feature selection is becoming more and more a challenging task due to the increase of the dimensionality of the data. The complexity of the interactions among features and the size of the search space make it unfeasible to find the optimal subset of features. In order to reduce the search space, feature grouping has arisen as an approach that allows to cluster feature according to the shared information about the class. On the other hand, metaheuristic algorithms have proven to achieve sub-optimal solutions within a reasonable time. In this work we propose a Scatter Search (SS) strategy that uses feature grouping to generate an initial population comprised of diverse and high quality solutions. Solutions are then evolved by applying random mechanisms in combination with the feature group structure, with the objective of maintaining during the search a population of good and, at the same time, as diverse as possible solutions. Not only does the proposed strategy provide the best subset of features found but it also reduces the redundancy structure of the data. We test the strategy on high dimensional data from biomedical and text-mining domains. The results are compared with those obtained by other adaptations of SS and other popular strategies. Results show that the proposed strategy can find, on average, the smallest subsets of features without degrading the performance of the classifier.
Gaia Collaboration
Gaia Data Release 3: Mapping the asymmetric disc of the Milky Way Journal Article
In: Astronomy and Astrophysics, 2023.
Abstract | Links | BibTeX | Tags: astrostatistics
@article{collaboration2022gaia,
title = {Gaia Data Release 3: Mapping the asymmetric disc of the Milky Way},
author = {Gaia Collaboration},
url = {https://www.aanda.org/component/article?access=doi&doi=10.1051/0004-6361/202243797},
doi = {10.1051/0004-6361/202243797},
year = {2023},
date = {2023-01-01},
journal = {Astronomy and Astrophysics},
abstract = {With the most recent Gaia data release the number of sources with complete 6D phase space information (position and velocity) has increased to well over 33 million stars, while stellar astrophysical parameters are provided for more than 470 million sources, in addition to the identification of over 11 million variable stars. Using the astrophysical parameters and variability classifications provided in Gaia DR3, we select various stellar populations to explore and identify non-axisymmetric features in the disc of the Milky Way in both configuration and velocity space. Using more about 580 thousand sources identified as hot OB stars, together with 988 known open clusters younger than 100 million years, we map the spiral structure associated with star formation 4-5 kpc from the Sun. We select over 2800 Classical Cepheids younger than 200 million years, which show spiral features extending as far as 10 kpc from the Sun in the outer disc. We also identify more than 8.7 million sources on the red giant branch (RGB), of which 5.7 million have line-of-sight velocities, allowing the velocity field of the Milky Way to be mapped as far as 8 kpc from the Sun, including the inner disc. The spiral structure revealed by the young populations is consistent with recent results using Gaia EDR3 astrometry and source lists based on near infrared photometry, showing the Local (Orion) arm to be at least 8 kpc long, and an outer arm consistent with what is seen in HI surveys, which seems to be a continuation of the Perseus arm into the third quadrant. Meanwhile, the subset of RGB stars with velocities clearly reveals the large scale kinematic signature of the bar in the inner disc, as well as evidence of streaming motions in the outer disc that might be associated with spiral arms or bar resonances. (abridged)},
keywords = {astrostatistics},
pubstate = {accepted},
tppubtype = {article}
}
With the most recent Gaia data release the number of sources with complete 6D phase space information (position and velocity) has increased to well over 33 million stars, while stellar astrophysical parameters are provided for more than 470 million sources, in addition to the identification of over 11 million variable stars. Using the astrophysical parameters and variability classifications provided in Gaia DR3, we select various stellar populations to explore and identify non-axisymmetric features in the disc of the Milky Way in both configuration and velocity space. Using more about 580 thousand sources identified as hot OB stars, together with 988 known open clusters younger than 100 million years, we map the spiral structure associated with star formation 4-5 kpc from the Sun. We select over 2800 Classical Cepheids younger than 200 million years, which show spiral features extending as far as 10 kpc from the Sun in the outer disc. We also identify more than 8.7 million sources on the red giant branch (RGB), of which 5.7 million have line-of-sight velocities, allowing the velocity field of the Milky Way to be mapped as far as 8 kpc from the Sun, including the inner disc. The spiral structure revealed by the young populations is consistent with recent results using Gaia EDR3 astrometry and source lists based on near infrared photometry, showing the Local (Orion) arm to be at least 8 kpc long, and an outer arm consistent with what is seen in HI surveys, which seems to be a continuation of the Perseus arm into the third quadrant. Meanwhile, the subset of RGB stars with velocities clearly reveals the large scale kinematic signature of the bar in the inner disc, as well as evidence of streaming motions in the outer disc that might be associated with spiral arms or bar resonances. (abridged)
L. Melgar-García and D. Gutiérrez-Avilés and C. Rubio-Escudero and A.Troncoso
Identifying novelties and anomalies for incremental learning in streaming time series forecasting Journal Article
In: Engineering Applications of Artificial Intelligence, vol. 123, pp. 106326, 2023.
Links | BibTeX | Tags: energy, IoT, time series
@article{Melgar2023b,
title = {Identifying novelties and anomalies for incremental learning in streaming time series forecasting},
author = {L. Melgar-García and D. Gutiérrez-Avilés and C. Rubio-Escudero and A.Troncoso},
url = {https://www.sciencedirect.com/science/article/pii/S0952197623005109},
doi = {https://doi.org/10.1016/j.engappai.2023.106326},
year = {2023},
date = {2023-01-01},
journal = {Engineering Applications of Artificial Intelligence},
volume = {123},
pages = {106326},
keywords = {energy, IoT, time series},
pubstate = {published},
tppubtype = {article}
}
L. Melgar-García and D. Gutiérrez-Avilés and C. Rubio-Escudero and A. Troncoso
A novel distributed forecasting method based on information fusion and incremental learning for streaming time series Journal Article
In: Information Fusion, vol. 95, pp. 163-173, 2023.
Links | BibTeX | Tags: energy, IoT, time series
@article{Melgar2023a,
title = {A novel distributed forecasting method based on information fusion and incremental learning for streaming time series},
author = {L. Melgar-García and D. Gutiérrez-Avilés and C. Rubio-Escudero and A. Troncoso},
url = {https://www.sciencedirect.com/science/article/pii/S1566253523000635},
doi = {https://doi.org/10.1016/j.inffus.2023.02.023},
year = {2023},
date = {2023-01-01},
journal = {Information Fusion},
volume = {95},
pages = {163-173},
keywords = {energy, IoT, time series},
pubstate = {published},
tppubtype = {article}
}
A. M. Chacón-Maldonado and G. Asencio-Cortés and F. Martínez-Álvarez and A. Troncoso
FS-Studio: An extensive and efficient feature selection experimentation tool for Weka Explorer Journal Article
In: SoftwareX, vol. 23, pp. 101401, 2023.
Links | BibTeX | Tags: feature selection
@article{Chacon2023,
title = {FS-Studio: An extensive and efficient feature selection experimentation tool for Weka Explorer},
author = {A. M. Chacón-Maldonado and G. Asencio-Cortés and F. Martínez-Álvarez and A. Troncoso},
url = {https://www.sciencedirect.com/science/article/pii/S2352711023000973},
doi = {https://doi.org/10.1016/j.softx.2023.101401},
year = {2023},
date = {2023-01-01},
journal = {SoftwareX},
volume = {23},
pages = {101401},
keywords = {feature selection},
pubstate = {published},
tppubtype = {article}
}
A. R. Troncoso-García and m. Martínez-Ballesteros and F. Martínez-Álvarez and A. Troncoso
Evolutionary computation to explain deep learning models for time series forecasting Conference
SAC 38th Annual ACM Symposium on Applied Computing, 2023.
Links | BibTeX | Tags: deep learning, time series, XAI
@conference{SAC2023,
title = {Evolutionary computation to explain deep learning models for time series forecasting},
author = {A. R. Troncoso-García and m. Martínez-Ballesteros and F. Martínez-Álvarez and A. Troncoso},
url = {https://dl.acm.org/doi/abs/10.1145/3555776.3578994},
year = {2023},
date = {2023-01-01},
booktitle = {SAC 38th Annual ACM Symposium on Applied Computing},
keywords = {deep learning, time series, XAI},
pubstate = {published},
tppubtype = {conference}
}
A. R. Troncoso-García and M. Martínez-Ballesteros and F. Martínez-Álvarez and A. Troncoso
Deep Learning-Based Approach for Sleep Apnea Detection Using Physiological Signals Conference
IWANN International Work-conference on Artificial Intelligence, Lecture Notes in Computer Science 2023.
BibTeX | Tags: deep learning, feature selection, time series
@conference{IWANN2023,
title = {Deep Learning-Based Approach for Sleep Apnea Detection Using Physiological Signals},
author = {A. R. Troncoso-García and M. Martínez-Ballesteros and F. Martínez-Álvarez and A. Troncoso},
year = {2023},
date = {2023-01-01},
booktitle = {IWANN International Work-conference on Artificial Intelligence},
series = {Lecture Notes in Computer Science},
keywords = {deep learning, feature selection, time series},
pubstate = {published},
tppubtype = {conference}
}