research

My research focuses on developing AI systems that are both powerful and interpretable. I am particularly interested in bridging the gap between high-performing machine learning models and human-understandable decision-making processes.

Rule-Based Learning

I am deeply interested in rule-based learning systems that produce interpretable decision rules. Unlike black-box models, rule-based classifiers generate explicit IF-THEN rules that practitioners can examine, validate, and trust.

My work in this area focuses on continuous optimization techniques for learning optimal rule bases. Traditional approaches often rely on combinatorial search, but continuous formulations enable the use of gradient-based methods and evolutionary algorithms to find better solutions more efficiently.

Key aspects of my work:

• Genetic algorithms for fuzzy rule optimization
• Differentiable rule learning approaches
• Balancing rule accuracy with interpretability
• Penalty-based aggregation for rule combination

Related publications:

2024

1. Interpreting contrastive embeddings in specific domains with fuzzy rules

   Javier Fumanal-Idocin ,  Mohammadreza Jamalifard ,  and  Javier Andreu-Perez

   In 2024 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) , 2024

Explainable Classification

A central theme in my research is explainable classification: building classifiers that not only achieve high accuracy but also provide meaningful explanations for their predictions.

I believe that for AI to be deployed responsibly in critical domains like healthcare, law, and security, we need models that humans can understand and verify. My work on Ex-Fuzzy directly addresses this need by providing a toolkit for creating explainable fuzzy classifiers.

Research directions:

• Fuzzy rule-based classifiers with linguistic interpretability
• Post-hoc explanation methods for deep learning
• Ensemble methods with explainable components (e.g., the Krypteia ensemble)
• Visual explanations for image classification

Related publications:

2025

1. Dynamic Feature Selection based on Rule-based Learning for Explainable Classification with Uncertainty Quantification

   Javier Fumanal-Idocin ,  Raquel Fernandez-Peralta ,  and  Javier Andreu-Perez

   arXiv preprint arXiv:2508.02566, 2025

   Abs arXiv

   Dynamic feature selection (DFS) offers a compelling alternative to traditional, static feature selection by adapting the selected features to each individual sample. This provides insights into the decision-making process for each case, which makes DFS especially significant in settings where decision transparency is key, i.e., clinical decisions. However, existing DFS methods use opaque models.

2023

1. 

   Artxai: Explainable artificial intelligence curates deep representation learning for artistic images using fuzzy techniques

   Javier Fumanal-Idocin ,  Javier Andreu-Perez ,  Oscar Cord , and 3 more authors

   IEEE Transactions on Fuzzy Systems, 2023

   arXiv

Uncertainty Quantification with Incomplete Information

Real-world data is often incomplete, imprecise, or affected by various sources of uncertainty. My research addresses how to make reliable decisions when working with such data.

I am particularly interested in:

• Interval-valued data: When measurements have inherent bounds of uncertainty
• Fuzzy representations: When categories have gradual rather than sharp boundaries
• Aggregation under uncertainty: Combining multiple uncertain sources of information

This work has applications in brain-computer interfaces, medical diagnosis, and any domain where data quality varies.

Related publications:

2026

1. Efficient online generation of fuzzy measures via aggregation functions

   Xabier Gonzalez-Garcia ,  L’ubomı́ra Horanská ,  Gleb Beliakov , and 1 more author

   Information Fusion, 2026

2025

1. Reliable Classification with Conformal Learning and Interval-Type 2 Fuzzy Sets

   Javier Fumanal-Idocin ,  and  Javier Andreu-Perez

   In 2025 IEEE International Conference on Fuzzy Systems (FUZZ) , 2025

   Abs arXiv

   Classical machine learning classifiers tend to be overconfident can be unreliable outside of the laboratory benchmarks. Properly assessing the reliability of the output of the model per sample is instrumental for real-life scenarios where these systems are deployed.

2. Reliable Classification with Conformal Learning and Interval-Type 2 Fuzzy Sets

   Javier Fumanal-Idocin ,  and  Javier Andreu-Perez

   In 2025 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) , 2025

2022

1. Almost aggregations in the gravitational clustering to perform anomaly detection

   J Fumanal-Idocin ,  I Rodriguez-Martinez ,  A Indurain , and 2 more authors

   Information Sciences, 2022

2021

1. Interval-Valued Aggregation Functions Based on Moderate Deviations Applied to Motor-Imagery-Based Brain–Computer Interface

   Javier Fumanal-Idocin ,  Zdenko Takáč ,  Javier Fernández , and 5 more authors

   IEEE Transactions on Fuzzy Systems, 2021
2. Motor-imagery-based brain–computer interface using signal derivation and aggregation functions

   Javier Fumanal-Idocin ,  Yu-Kai Wang ,  Chin-Teng Lin , and 3 more authors

   IEEE Transactions on Cybernetics, 2021

2019

1. Gravitational clustering algorithm generalization by using an aggregation of masses in newton law

   J Armentia ,  Iosu Rodrı́guez ,  Javier Fumanal Idocin , and 3 more authors

   In New Trends in Aggregation Theory 10 , 2019

Fuzzy Logic

Fuzzy logic provides the mathematical foundation for much of my work. It allows us to model imprecise concepts using degrees of membership rather than binary true/false values, making it particularly suited for human-centric AI applications.

My contributions to fuzzy logic include:

• Fuzzy integrals: Generalizations of the Sugeno and Choquet integrals for data fusion
• Fuzzy clustering: Using fuzzy membership for community detection and anomaly detection
• Interval-valued fuzzy sets: Extensions that capture additional uncertainty in membership degrees
• Type-2 fuzzy systems: Handling uncertainty about the fuzzy sets themselves

I maintain Ex-Fuzzy, a Python library that makes fuzzy logic accessible to the broader machine learning community.

Related publications:

2026

1. Efficient online generation of fuzzy measures via aggregation functions

   Xabier Gonzalez-Garcia ,  L’ubomı́ra Horanská ,  Gleb Beliakov , and 1 more author

   Information Fusion, 2026

2025

1. Crisp complexity of fuzzy classifiers

   Raquel Fernandez-Peralta ,  Javier Fumanal-Idocin ,  and  Javier Andreu-Perez

   In 2025 IEEE International Conference on Fuzzy Systems (FUZZ) , 2025

   Abs arXiv

   Rule-based systems are a very popular form of explainable AI, particularly in the fuzzy community, where fuzzy rules are widely used for control and classification problems.

2. Reliable Classification with Conformal Learning and Interval-Type 2 Fuzzy Sets

   Javier Fumanal-Idocin ,  and  Javier Andreu-Perez

   In 2025 IEEE International Conference on Fuzzy Systems (FUZZ) , 2025

   Abs arXiv

   Classical machine learning classifiers tend to be overconfident can be unreliable outside of the laboratory benchmarks. Properly assessing the reliability of the output of the model per sample is instrumental for real-life scenarios where these systems are deployed.

3. Crisp complexity of fuzzy classifiers

   Raquel Fernandez-Peralta ,  Javier Fumanal-Idocin ,  and  Javier Andreu-Perez

   In 2025 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) , 2025
4. Reliable Classification with Conformal Learning and Interval-Type 2 Fuzzy Sets

   Javier Fumanal-Idocin ,  and  Javier Andreu-Perez

   In 2025 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) , 2025
5. A Fast Interpretable Fuzzy Tree Learner

   Javier Fumanal-Idocin ,  Raquel Fernandez-Peralta ,  and  Javier Andreu-Perez

   arXiv preprint arXiv:2512.11616, 2025

   arXiv

2024

1. 

   Ex-Fuzzy: A library for symbolic explainable AI through fuzzy logic programming

   Javier Fumanal-Idocin ,  and  Javier Andreu-Perez

   Neurocomputing, 2024

   Abs HTML Code

   Understanding the decisions taken by machine learning systems is instrumental in their deployment in real-world systems, as it enables responsible decision-making, fosters trust, and facilitates debugging and improvement. The research field devoted to studying the techniques that explain and illustrate those decisions is called explainable AI. Fuzzy logic, with its interpretable fuzzy rule-based inference, has emerged as a popular tool for Explainable AI because of these interpretable classifiers. However, current fuzzy logic libraries provide limited inference capabilities and integration to machine learning or are only available in the Java or R language, which makes their integration with the standard machine libraries in Python challenging. This paper describes a software library that contains a Python implementation to perform fuzzy inference using different kinds of fuzzy sets, with a special focus on result visualization. This library follows the scikit-learn programming interface, enabling researchers to utilize it with minimum fuzzy logic background seamlessly. This toolkit unveils novel tools for programming fuzzy systems that are learnable using machine learning methods, leading to data-powered systems that maintain full transparency and accountability, accessible to virtually anyone without specialized AI training. The interpretability of these systems makes them highly valuable in industries like healthcare, law, and security.

2. Interpreting contrastive embeddings in specific domains with fuzzy rules

   Javier Fumanal-Idocin ,  Mohammadreza Jamalifard ,  and  Javier Andreu-Perez

   In 2024 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) , 2024

2023

1. 

   Artxai: Explainable artificial intelligence curates deep representation learning for artistic images using fuzzy techniques

   Javier Fumanal-Idocin ,  Javier Andreu-Perez ,  Oscar Cord , and 3 more authors

   IEEE Transactions on Fuzzy Systems, 2023

   arXiv
2. On the Stability of Fuzzy Classifiers to Noise Induction

   Javier Fumanal-Idocin ,  Humberto Bustince ,  Javier Andreu-Perez , and 1 more author

   In 2023 IEEE International Conference on Fuzzy Systems (FUZZ) , 2023

2022

1. Fuzzy Clustering to Encode Contextual Information in Artistic Image Classification

   Javier Fumanal-Idocin ,  Zdenko Takáč ,  L’ubomı́ra Horanská , and 2 more authors

   In International Conference on Information Processing and Management of Uncertainty in Knowledge-Based Systems , 2022

2020

1. Adaptive binarization based on fuzzy integrals

   Francesco Bardozzo ,  Borja De La Osa ,  Lubomira Horanska , and 6 more authors

   arXiv preprint arXiv:2003.08755, 2020

2019

1. Distances between interval-valued fuzzy sets taking into account the width of the intervals

   Zdenko Takáč ,  Javier Fernandez ,  Javier Fumanal , and 5 more authors

   In 2019 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) , 2019

Collaboration

I am always open to collaborations on these topics. If you are interested in working together or have questions about my research, please feel free to reach out.