Pasar al contenido principal

APRENDIZAJE AUTOMATICO FRUGAL: POTENCIANDO LA IA EN ENTORNOS CON RECURSOS LIMITADOS PARA LOS DESAFIOS DEL MUNDO REAL

PID2023-147404OB-I00

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia
Subprograma Subprograma Estatal de Generación de Conocimiento
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2023
Plan Estatal 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSIDADE DA CORUÑA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 13
Encontrada(s) 1 página(s)

RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Improving restaurant recommendation transparency through feature selection

  • Bagué-Masanés, Roger
  • Remeseiro, Beatriz
  • Bolón-Canedo, Verónica
[Abstract]: Recommender systems are widely used to suggest items (i.e., products or services) based on user preferences. However, personalized recommender systems incorporating information beyond user ratings can provide more accurate and relevant recommendations. This paper introduces a hybrid personalized recommender system that utilizes several characteristics of items in addition to user ratings. The focus of this study is on the restaurant industry, and the dataset used is sourced from TripAdvisor, one of the most popular travel and tourism websites. The attributes include price range, cuisines, special diets, meals, and features like free WiFi and wheelchair accessibility. To improve the transparency of the recommendation process and understand the variables that influence the system’s output, we propose the use of feature selection techniques. By analyzing the impact of each variable, this study aims to help readers understand the recommendation process and identify the factors to consider when choosing to visit a restaurant., This work has been supported by the National Plan for Scientific and Technical Research and Innovation of the Spanish Government (Grants PID2023-147404OB-I00 and TED2021-130599A-I00), and by the Ministry for Digital Transformation and Civil Service and ‘Next-GenerationEU’/PRTR under Grant TSI-100925-2023-1. The Agency for Science, Business Competitiveness, and Innovation of the Principality of Asturias in Spain (SEKUENS) is also acknowledged for funding through the project GRU-GIC-24–018. CITIC, as a center accredited for excellence within the Galician University System and a member of the CIGUS Network, receives subsidies from the Department of Education, Science, Universities, and Vocational Training of the Xunta de Galicia. Additionally, it is co-financed by the EU through the FEDER Galicia 2021–27 operational program (Ref. ED431G 2023/01). Grant ED431C 2022/44 funded by, Xunta de Galicia; ED431G 2023/01, Xunta de Galicia; ED431C 2022/44, Gobierno del Principado de Asturias; GRU-GIC-24–018
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Enhancing Classification Performance on Imbalanced Datasets Through Complexity-Guided Oversampling With SMOTE

  • Morillo-Salas, José Luis
  • Bolón-Canedo, Verónica
  • Morán-Fernández, Laura
  • Alonso-Betanzos, Amparo
[Abstract]: Improving classification performance on imbalanced datasets remains a challenging problem in machine learning. Synthetic oversampling techniques such as SMOTE are widely used to address class imbalance; however, their random interpolation strategy often ignores structural data properties, which may affect classifier generalisation. This work proposes a set of SMOTE-based strategies that guide the generation of synthetic samples in order to produce structurally simpler training datasets that are easier for classifiers to learn. The first strategy generates more dispersed (outer) synthetic samples to increase class separability with minimal computational overhead. The second and main contribution, SMOTE-Complex, formulates synthetic sample selection as an explicit optimisation process that minimises measurable training dataset complexity. A clustering-based variant, SMOTE-Complex-Clustering, reduces computational cost by restricting optimisation to feature subspaces while preserving most structural and predictive benefits. The underlying hypothesis is that reducing the structural complexity of the training data can lead to improved predictive behaviour. Extensive experiments on binary and multiclass datasets, using multiple classifiers and complementary evaluation metrics, provide empirical support for this hypothesis across diverse structural conditions. The results indicate moderate but stable improvements in structurally favourable scenarios—particularly binary and moderately complex problems—without systematic degradation in imbalance-sensitive metrics, while the clustering-based refinement offers a scalable trade-off between optimisation strength and computational efficiency., This work has been supported by the National Plan for Scientific and Technical Research and Innovation of the Spanish Government (Grant PID2023-147404OB-I009), and by the Ministry for Digital Transformation and Civil Service and ‘Next-GenerationEU’/PRTR under Grant TSI-100925-2023-1. CITIC, as a center accredited for excellence within the Galician University System and a member of the CIGUS Network, receives subsidies from the Department of Education, Science, Universities, and Vocational Training of the Xunta de Galicia. Additionally, it is co-financed by the EU through the FEDER Galicia 2021-27 operational program (Ref. ED431G 2023/01). Grant ED431C 2022/44 funded by Xunta de Galicia., Xunta de Galicia; ED431G 2023/01, Xunta de Galicia; ED431C 2022/44
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Optimising Image Feature Extraction and Selection: A Comprehensive Review With Spark Case Studies

  • Figueira-Domínguez, J. Guzmán
  • Remeseiro, Beatriz
  • Bolón-Canedo, Verónica
The data that support the findings of this study are openly available inImagenet Features Extracted with VGG-19 at https://zenodo.org/records/12791398, [Abstract]: As benchmark image datasets expand in sample size and feature complexity, the challenge of managing increased dimensionality becomes apparent. Contrary to the expectation that more features equate to enhanced information and improved outcomes, the curse of dimensionality often hampers performance. This paper reviews existing literature on filter feature selection techniques applied to image features, highlighting their use in both classical and deep-learning-based feature extraction methods. Building on these findings, this study proposes a scalable approach for image feature extraction and selection using Big Data technologies, specifically Apache Spark, to efficiently process large and high-dimensional datasets. The proposed framework integrates filter-based feature selection methods within a distributed environment to evaluate their effectiveness in image analysis tasks. Several experiments were performed to compare the results using feature selection techniques with various reduction percentages. Results show that significant feature reduction can be achieved without compromising classification accuracy, demonstrating the potential of Spark-based distributed processing for large-scale image analytics., This work was supported by Xunta de Galicia (ED431C 2022/44);National Plan for Scientific and Technical Research and Innovation ofthe Spanish Government (PID2023-147404OB-I00); Ministry for DigitalTransformation and Civil Service and Next-GenerationEU/PRTR (TSI-100925-2023-1); Agency for Science, Business Competitiveness, and Innovation of the Principality of Asturias in Spain (SEKUENS) (GRU-GIC- 24-018); Department of Education, Science, Universities, andVocational Training of the Xunta de Galicia; FEDER Galicia 2021-27 op-erational program (ED431G 2023/01), Xunta de Galicia; ED431C 2022/44, Xunta de Galicia; ED431G 2023/01, Gobierno del Principado de Asturias; GRU-GIC- 24-018
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Contrastive Learning for Explanation Ranking

  • Escarda Fernández, Miguel
  • Cancela, Brais
  • Eiras-Franco, Carlos
  • Guijarro-Berdiñas, Bertha
  • Alonso-Betanzos, Amparo
Financiado para publicación en acceso aberto: CRUE-CSIC, [Abstract]: Explainable recommendation systems enhance user trust and satisfaction by revealing the reasoning behind personalized recommendations. Approaching this as a post-hoc explanation-ranking problem over a fixed pool of candidate explanations, we propose Contrastive Learning for Explanation Ranking (CLER), a model that learns user, item, and explanation representations with a Normalized Temperature-scaled Binary Cross-Entropy (NT-BXent) loss. This function specifically applies a per-row reweighting strategy, preventing the vast number of negative examples from dominating the objective. We evaluate CLER on the Amazon, TripAdvisor, and Yelp datasets from the EXTRA benchmark. Across traditional ranking metrics, CLER achieves the strongest results among the compared baselines., This work was supported in part by MICIU/AEI/10.13039/501100011033 and ERDF, EU under Grant PID2023-147404OB-I00 and Grant PID2021-128045OA-I00; in part by the Ministry for Digital Transformation and Civil Service and Next-GenerationEU/PRTR under Grant TSI-100925-2023-1; in part by the Xunta de Galicia under Grant ED431C 2022/44; and in part by Consellería de Cultura, Educación e Universidade (ERDF Operational Programme Galicia 2021–2027) and Secretaría Xeral de Universidades under Grant ED431G2023/01., Xunta de Galicia; ED431C 2022/44, Xunta de Galicia; ED431G2023/01
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Beyond RMSE and MAE: Introducing EAUC to Unmask Hidden Bias and Unfairness in Dyadic Regression Models

  • Paz Ruza, Jorge
  • Alonso-Betanzos, Amparo
  • Guijarro-Berdiñas, Bertha
  • Cancela, Brais
  • Eiras-Franco, Carlos
[Abstract]: Dyadic regression models, which output real-valued predictions for pairs of entities, are fundamental in many domains [e.g., obtaining user-product ratings in recommender systems (RSs)] and promising and under exploration in others (e.g., tuning patient–drug dosages in precision pharmacology). In this work, we prove that nonuniform observed value distributions of individual entities lead to severe biases in state-of-the-art models, skewing predictions toward the average of observed past values for the entity and providing worse-than-random predictive power in eccentric yet crucial cases; we name this phenomenon eccentricity bias. We show that global error metrics like root-mean-squared error (RMSE) are insufficient to capture this bias, and we introduce eccentricity area under the curve (EAUC) as a novel metric that can quantify it in all studied domains and models. We prove the intuitive interpretation of EAUC by experimenting with naive post-training bias corrections and theorize other options to use EAUC to guide the construction of fair models. This work contributes a bias-aware evaluation of dyadic regression to prevent unfairness in critical real-world applications of such systems., This work was supported in part by MICIU/AEI/10.13039/501100011033 and ESF+ under Grant FPU21/05783; in part by MICIU/AEI/10.13039/501100011033 under Grant PID2019-109238GB-C22; in part by MICIU/AEI/10.13039/501100011033 and ERDF, EU under Grant PID2023-147404OB-I00 and Grant PID2021-128045OA-I00; in part by the Ministry for Digital Transformation and Civil Service and Next-Generation EU/PRTR under Grant TSI-100925-2023-1; in part by the Xunta de Galicia under Grant ED431C 2022/44; and in part by Consellería de Cultura, Educación e Universidade (ERDF Operational Programme Galicia 2021–2027) and Secretaría Xeral de Universidades under Grant ED431G 2023/01., Xunta de Galicia; ED431C 2022/44, Xunta de Galicia; ED431G 2023/01
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

A Hybrid Metaheuristics-Bayesian Optimization Framework with Safe Transfer Learning for Continuous Spark Tuning

  • Garralda-Barrio, Mariano
  • Eiras-Franco, Carlos
  • Bolón-Canedo, Verónica
The source code and experimental resources supporting this study are openly available at Github (https://github.com/mgarralda/garralda-performance-model). They are released for academic and research use under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0)., [Abstract]: Tuning configuration parameters in distributed Big Data engines such as Apache Spark is a high-dimensional, workload-dependent problem with significant impact on performance and operational cost. We address this challenge with a hybrid optimization framework that integrates Iterated Local Search, Tabu Search, and locally embedded Bayesian Optimization guided by STL-PARN (safe transfer learning with pattern-adaptive robust neighborhoods). Historical executions are partitioned into a Nucleus of reliable neighbors and a Corona of exploratory configurations, ensuring relevance while mitigating negative transfer. The surrogate within the embedded Bayesian Optimization stage decouples performance prediction from uncertainty modeling, enabling parameter-free acquisition functions that self-adapt to diverse workloads. Experiments on a modernized HiBench suite across multiple input scales show consistent gains over state-of-the-art baselines in execution time, convergence, and cost efficiency. Overall, the results demonstrate the robustness and practical value of embedding Bayesian Optimization within a global metaheuristic loop for adaptive, cost-aware Spark tuning. All source code and datasets are publicly available, supporting reproducibility and operational efficiency in large-scale data processing., The authors extend their gratitude to Laboratorio Innovación Aplicada (LIA) at Minsait (Indra Company) for their support in this study. This work has been supported by Ministerio de Ciencia e Innovación MCIN/AEI/10.13039/501100011033/ FEDER, UE under grant PID2023-147404OB-I00, and by the Ministry for Digital Transformation and Civil Service and ‘Next-GenerationEU’ /PRTR under Grant TSI-100925-2023-1. CITIC, as a center accredited for excellence within the Galician University System and a member of the CIGUS Network, receives subsidies from the Department of Education, Science, Universities, and Vocational Training of the Xunta de Galicia. Additionally, it is co-financed by the EU through the FEDER Galicia 2021-27 operational program (Ref. ED431G 2023/01). Grant ED431C 2022/44 funded by Xunta de Galicia., Xunta de Galicia; ED431G 2023/01, Xunta de Galicia; ED431C 2022/44
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Machine Learning for Perovskite Solar Cells: A Comprehensive Review on Opportunities and Challenges for Materials Scientists

  • de la Asunción-Nadal, Víctor
  • Sprague, Christopher Iliffe
  • Guijarro-Berdiñas, Bertha
  • Cappel, Ute
  • García-Fernández, Alberto
[Abstract] Perovskite solar cells (PSCs) have emerged as a promising technology due to their tunable optoelectronic properties, low-cost fabrication and high efficiency. Despite this progress, key challenges such as long-term stability, large-scale manufacturability, and recyclability remain unsolved. Moreover, traditional methods for discovering new materials and optimizing device architectures rely on trial-and-error experiments. Machine learning (ML) offers powerful tools to address these bottlenecks by uncovering hidden patterns in data, accelerating discovery, and guiding rational design. However, the growing number of ML-driven studies in PSC research can be difficult to navigate, particularly for experimentalists and scientists without a computational background. To address this gap, this review is written with accessibility in mind, and it is structured to serve as a bridge between ML experts and the broader materials science community. We provide an overview of how ML can be applied to PSCs, from databases and data preprocessing to model training, evaluation, and interpretability. Advantages and limitations of different approaches are critically assessed, with emphasis on how dataset choice, algorithms, and metrics affect reliability. We conclude by outlining current challenges and open questions, as well as potential directions where the integration of ML with experimental and theoretical research could further advance the development of perovskite solar cells., We thank the Swedish Research Council (Grant No. VR 2022-03168) and the Göran Gustafsson foundation for funding. This work was partially supported by the Wallenberg Initiative Materials Science for Sustainability (WISE) funded by the Knut and Alice Wallenberg Foundation. We also acknowledge Project PID2023-147404OB-I00 funded by MCIN/AEI/10.13039/501100011033/ERDF, UE, and Xunta de Galicia (Grant ED431C 2022/44). A. G.-F. acknowledges support from a Beatriz Galindo junior fellowship (BG23/00033) from the Spanish Ministry of Science and Innovation, Suecia. Vetenskapsrådet; 2022-03168, Xunta de Galicia; ED431C 2022/44
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Adaptive incremental transfer learning for efficient performance modeling of big data workloads

  • Garralda-Barrio, Mariano
  • Eiras-Franco, Carlos
  • Bolón-Canedo, Verónica
This manuscript version is made available under the CC-BY-NC-ND 4.0 license https://creativecommons.org/licenses/bync-nd/4.0/. This version of the article has been accepted for publication in Future Generation Computer Systems. The Version of Record is available online at https://doi.org/10.1016/j.future.2025.107730, The algorithms, evaluation metrics, cross-validation methods, and resources used in this study are openly available at https://github.com/mgarralda/garralda-performance-model. Further details can be obtained from the corresponding author on reasonable request., [Abstract]: The rise of data-intensive scalable computing systems, such as Apache Spark, has transformed data processing by enabling the efficient manipulation of large datasets across machine clusters. However, system configuration to optimize performance remains a challenge. This paper introduces an adaptive incremental transfer learning approach to predicting workload execution times. By integrating both unsupervised and supervised learning, we develop models that adapt incrementally to new workloads and configurations. To guide the optimal selection of relevant workloads, the model employs the coefficient of distance variation (CdV) and the coefficient of quality correlation (CqC), combined in the exploration–exploitation balance coefficient (EEBC). Comprehensive evaluations demonstrate the robustness and reliability of our model for performance modeling in Spark applications, with average improvements of up to 31% over state-of-the-art methods. This research contributes to efficient performance tuning systems by enabling transfer learning from historical workloads to new, previously unseen workloads. The full source code is openly available., This work has been supported by the National Plan for Scientific and Technical Research and Innovation of the Spanish Government, Spain (Grant PID2019-109238GB-C22 and PID2023-147404OB-I00), and by the Xunta de Galicia (Grant ED431C 2022/44) with the European Union ERDF funds, Spain. CITIC, as Research Center accredited by Galician University System, is funded by “Consellería de Cultura, Educación e Universidade from Xunta de Galicia, Spain ”, supported in an 80% through ERDF Operational Programme Galicia 2014–2020, and the remaining 20% by “Secretaría Xeral de Universidades” (Grant ED431G 2023/01)., Xunta de Galicia; ED431C 2022/44, Xunta de Galicia; ED431G 2023/01
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Minimizing Redundancy in Hand Dynamic Features for Enhanced Sign Language Recognition

  • Abdullahi, Sunusi Bala
  • Bolón-Canedo, Verónica
Versión aceptada de: S. Bala Abdullahia, and V. Bolon-Canedo, "Minimizing Redundancy in Hand Dynamic Features for Enhanced Sign Language Recognition", Intelligent Data Analysis: An International Journal, Vol. 30, Issue 3. Copyright © 2025 SAGE. DOI: 10.1177/1088467X251367228., [Abstract]: This paper presents a wavelet-based framework for 3-D sign language recognition that outperforms conventional ZCA approaches by preserving critical spatiotemporal relationships in hand kinematics. We formulate hand dynamics as a wavelet exponent energy problem using Daubechies transforms, where joint coordinates, mother coefficients, and window sizes are jointly optimized, avoiding PCA’s disruptive axis rotation while maintaining interpretable feature orientations. The proposed method combines this wavelet energy representation with an ensemble deep learning feature selection layer using binary masking (λ1 = 0.10, λ2 = 0.01) and regularization to eliminate redundancy, achieving 94.09% accuracy (vs. 92.75% for PCC+ZCA (SVD)) with 4.3× faster inference on Turkish SL datasets. Statistical validation (ANOVA, p < 0.001) confirms significant improvements over ZCA-derived features (t = 24.15), with weight analysis revealing proximal arm kinematics as dominant recognition factors. The framework’s computational efficiency (0.12ms inference) and preservation of kinematic relationships demonstrate its superiority for real-life SLR applications compared to both PCC and ZCA approaches., This work has been supported by the National Plan for Scientific and Technical Research and Innovation of the Spanish Government (Grant PID2023-147404OB-I00), and by the Ministry for Digital Transformation and Civil Service and ‘Next-GenerationEU’/PRTR under Grant TSI-100925-2023-1. CITIC, as a center accredited for excellence within the Galician University System and a member of the CIGUS Network, receives subsidies from the Department of Education, Science, Universities, and Vocational Training of the Xunta de Galicia. Additionally, it is co-financed by the EU through the FEDER Galicia 2021-27 operational program (Ref. ED431G 2023/01). Grant ED431C 2022/44 funded by Xunta de Galicia., Xunta de Galicia; ED431G 2023/01, Xunta de Galicia; ED431C 2022/44
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Performance and sustainability of BERT derivatives in dyadic data

  • Escarda Fernández, Miguel
  • Eiras-Franco, Carlos
  • Cancela, Brais
  • Guijarro-Berdiñas, Bertha
  • Alonso-Betanzos, Amparo
Financiado para publicación en acceso aberto: Universidade da Coruña/CISUG, [Abstract]: In recent years, the Natural Language Processing (NLP) field has experienced a revolution, where numerous models – based on the Transformer architecture – have emerged to process the ever-growing volume of online text-generated data. This architecture has been the basis for the rise of Large Language Models (LLMs). Enabling their application to many diverse tasks in which they excel with just a fine-tuning process that comes right after a vast pre-training phase. However, their sustainability can often be overlooked, especially regarding computational and environmental costs. Our research aims to compare various BERT derivatives in the context of a dyadic data task while also drawing attention to the growing need for sustainable AI solutions. To this end, we utilize a selection of transformer models in an explainable recommendation setting, modeled as a multi-label classification task originating from a social network context, where users, restaurants, and reviews interact., Research funded by the Spain Ministry of Science, Innovation and Universities (MCIN/AEI/10.13039/501100011033) and ERDF “A way of making Europe” (PID2019-109238GB-C22, PID2021-128045OA-I00, PID2023-147404OB-I00), by the “Xunta de Galicia” (ED431C 2022/44) with the European Union ERDF funds, and by the Ministry for Digital Transformation and Civil Service and “Next-GenerationEU”/PRTR (Grant TSI-100925-2023-1). CITIC, as Research Center accredited by Galician University System, is funded by “Consellería de Cultura, Educación e Universidade from Xunta de Galicia ”, supported in an 80% through ERDF Operational Programme Galicia 2014-2020, and the remaining 20% by “Secretaría Xeral de Universidades” (Grant ED431G 2023/01)., Xunta de Galicia; ED431C 2022/44, Xunta de Galicia; ED431G 2023/01
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Breaking Boundaries: Low-Precision Conditional Mutual Information for Efficient Feature Selection

  • Morán-Fernández, Laura
  • Blanco-Mallo, Eva
  • Sechidis, Konstantinos
  • Bolón-Canedo, Verónica
[Abstract] As internet-of-things (IoT) devices proliferate, the need for efficient data processing at the network edge becomes increasingly critical due to the vast amounts of data generated. This paper presents a groundbreaking approach that leverages edge computing to address these challenges, using low-precision conditional mutual information (CMI) for feature selection. Our novel methodology improves the efficiency of edge computing systems by significantly reducing memory and energy consumption while maintaining high accuracy. We adapt this approach to feature selection algorithms, specifically, conditional mutual information maximization (CMIM) and incremental association Markov blanket (IAMB), and demonstrate its effectiveness for diverse datasets, including complex DNA microarrays. Our results show that low-precision methods not only compare competitively with traditional 64-bit implementations, but also yield significant performance and resource savings. For IoT and other machine learning applications, this work represents a significant advance in the development of more sustainable and efficient algorithms that can optimize computational resources and reduce their environmental impact., This work was supported in part by Xunta de Galicia/FEDER-UE under Grant ED431G 2023/01; Ministerio de Ciencia e Innovacion MCIN/AEI/10.13039/501100011033 and ‘Next-GenerationEU’/PRTR under Grants [PID2019-109238GB-C22; TED2021-130599A-I00; PID2023-147404OB-I00] and Ministry for Digital Transformation and Civil Service and ‘Next-GenerationEU’/PRTR under grant TSI-100925-2023-1. CITIC, as a Research Center of the University System of Galicia, is funded by Consellería de Educacion, Universidade e Formación Profesional of the Xunta de Galicia, Spain through the European Regional Development Fund (ERDF/FEDER) and the Secretaría Xeral de Universidades (ED431G 2023/01)., Xunta de Galicia; ED431G 2023/01
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Efficient Single-Step Framework for Incremental Class Learning in Neural Networks

  • Dopico Castro, Alejandro
[Abstract]: Incremental learning continues to present a significant challenge in deep learning, particularly in environments where resources are limited. While existing methods have been shown to achieve high levels of accuracy, they often require substantial computational resources and storage capacity. This work proposes CIFNet, an efficient approach to class incremental learning that matches comparable accuracy to state-of-the-art methods while significantly reducing training time and energy consumption through a single-step optimisation process. CIFNet incorporates a novel compressed buffer mechanism that stores condensed representations of previous data samples instead of full raw data, substantially reducing memory requirements. In contrast to conventional approaches that necessitate multiple iterations of weight optimisation, our method achieves optimal performance in a single training step, with no need for iterations, leading to a significant decrease in computational overhead. Experimental results on standard benchmark datasets have shown that our approach inherently mitigates catastrophic forgetting without the need for complex regularization schemes. CIFNet achieves accuracy comparable to current state-of-the-art approaches while significantly reducing training time and energy consumption. This work represents a step forward in making class incremental learning more accessible for resource-constrained environments while maintaining robust performance., This work was partially funded by Horizon Europe, GA 101070381 (‘PILLAR-Robots - Purposeful Intrinsically motivated Lifelong Learning Autonomous Robots’); Project PID2023-147404OB-I00 funded by MCIN/AEI/10.13039/501100011033/ ERDF, UE; Ministry for Digital Trans- formation and Civil Service and ‘Next-GenerationEU’/PRTR under Grant TSI-100925-2023-1; and Xunta de Galicia/FEDER-UE under Grant ED431C 2022/44. CITIC, as Research Center accredited by Galician University System, is funded by “Consellería de Cultura, Educación e Universidade from Xunta de Galicia”, supported in an 80% through ERDF Operational Programme Galicia 2014-2020, and the remaining 20% by “Secretaría Xeral de Universidades” (Grant ED431G 2023/01)., Xunta de Galicia; ED431C 2022/44, Xunta de Galicia; ED431G 2023/01, Traballo fin de mestrado (UDC.FIC). Intelixencia Artificial. Curso 2024/2025
RUC. Repositorio da Universidade da Coruña
Publicación Artículo científico (JournalArticle)

Optimising resource use through low-precision feature selection: a performance analysis of logarithmic division and stochastic rounding

  • Suárez-Marcote, Samuel
  • Morán-Fernández, Laura
  • Bolón-Canedo, Verónica
This is the peer reviewed version of the article which has been published in final form at https://doi.org/10.1111/exsy.70012., [Abstract]: The growth in the number of wearable devices has increased the amount of data produced daily. Simultaneously, the limitations of such devices has also led to a growing interest in the implementation of machine learning algorithms with low-precision computation. We propose green and efficient modifications of state-of-the-art feature selection methods based on information theory and fixed-point representation. We tested two potential improvements: stochastic rounding to prevent information loss, and logarithmic division to improve computational and energy efficiency. Experiments with several datasets showed comparable results to baseline methods, with minimal information loss in both feature selection and subsequent classification steps. Our low-precision approach proved viable even for complex datasets like microarrays, making it suitable for energy-efficient internet-of-things (IoT) devices. While further investigation into stochastic rounding did not yield significant improvements, the use of logarithmic division for probability approximation showed promising results without compromising classification performance. Our findings offer valuable insights into resource-efficient feature selection that contribute to IoT device performance and sustainability., This work was supported by CITIC, a Research Center accredited by Galician University System, which is funded by ‘Consellería de Cultura, Educación e Universidade from Xunta de Galicia’, supported in an 80% through ERDF Funds, ERDF Operational Programme Galicia 2014–2020, and the remaining 20% by ‘Secretaría Xeral de Universidades, Xunta de Galicia’ (Grant ED431G 2023/01). It was also partially funded by Xunta de Galicia/FEDER-UE under Grant ED431C 2022/44; Ministerio de Ciencia, Innovación y Universidades MCIN/AEI/10.13039/501100011033 and ‘Next-GenerationEU’/PRTR under Grants (PID2023-147404OB-I00; TED2021-130599A-I00) and Ministry for Digital Transformation and Civil Service under grant TSI-100925-2023-1. It was also supported by the Programa de axudas á etapa predoutoral of the Consellería de Cultura, Educación, Universidade e Formación Profesional, Xunta de Galicia, Spain (Ref. ED481A 2023/034)., Xunta de Galicia; ED431G 2023/01, Xunta de Galicia; ED431C 2022/44, Xunta de Galicia; ED481A 2023/034