Enrique Valero-Leal
Researcher, instructor and more
A PhD in AI candidate and software developer that cares about the ethical challenges that new technologies pose to the modern world.
Hello there! My name is
About Me
I am currently a pre-doctoral researcher at the CIG lab in the technical university of Madrid, where I research about robust and responsible artificial intelligence. My passion is to study from different angles the challenges that we are confronted with in the modern world and to tackle them with the tools and knowledge that I acquired during my whole career.
I have a wide academic background, having worked in Murcia, Tokyo and currently Madrid, and explored many fields, such as healthcare and education. Even in Madrid, I often participate in congresses around the world and work alongside professors from other universities, such as from Utrecht, Netherlands. Regarding industry, I currently collaborate with entrerprises in R&D projects about industry 4.0
Here is a list of technologies and skills that are present in my work- Explainable AI
- Probabilistic models
- Neural networks
- PyTorch
- Statistics
- MLOps
Experience
I started working as a researcher in AI in 2021 at Universidad Politécnica de Madrid and since 2022 I am a predoctoral student in AI as well. I develop cutting-edge projects related with interpreting probabilistic graphical models and where collaborations with other institutions and industry are present. Some examples are cooperation with Utrecht university and Repsol. I usually work with:
- Probabilistic artificial intelligence
- Explainable artificial intelligence
- Software development in HPC and distributed systems
- Basic MLOps
Find me among its members
Instructor of the course on explainable AI of the Machine Learning and Advanced Statistics Summer School. The course manages to attract the most student out of all them and receives some of the best feedback from students. Some of the topocs of the course are:
- Introduction of explainable AI
- Model agnostic explanations: Feature contribution, counterfactuals, SHAP…
- Interpretation of neural networks
- Causality and causal models
I collaborate in R&D related tasks in Repsol Technology Lab projects.
Currently my main tasks focus on supervising the deployment of a screening expert model for energy efficiency.
I did a research visit to the Tokyo Institute of technology, where I worked in AI in Education within the CrossLab, an elite multidisciplinary research group that combines the best work practices from Japan and USA.
In particular, the project that I developed evaluated which decisions in the learning process of a student had the bigger impact into his knowledge mastery. I worked with
- Domain specific literature
- Recurrent neural networks
At a more personal level, it was an experienced that enriched my perception of the academic world.
Find me in Lab alumni section!
My whole academy career started in here, in the University of my hometown. I was granted a state fellowship for novel researchers that allow to research in the field of pattern mining and explainable AI within the Department of Information and Communication Engineering, specifically in tasks of the AIKE group
Education
2022 - currently
PhD in Artificial Intelligence: Bayesian networks for Machine Learning Interpretability
Universidad Politécnica de Madrid
Advanced research on how to use Bayesian networks to improve significantly explainability of machine learning models.
The objective of this thesis is to tackle the wide variety of challenges that arise from the use of complex artificial intelligences in more and more aspects of our lifes. Currently the main points of research are:
- Feature importance of both observed and unobserved variables
- Counterfactual explanations
- Argumentation frameworks
2020 - 2022
MSc in Artificial Intelligence
Universidad Politécnica de Madrid
GPA:
9.64 out of 10. Highest grade of the promotion
Specialization in neural networks and optimization, with honorary mention in machine learning and Bayesian networks.
Extracurricular activities
- STEM mentor in low income high schools
2016 - 2020
BSc in Computer Science
University of Murcia
GPA:
9.4 out of 10. Highest grade of the promotion
Specialization in Artificial Intelligence and Computing, with honorary mention in +25 subjects out of 40
Extracurricular Activities
- Research in various departments, 2017-2020
- State fellowship for young researchers, 2019/20
- Erasmus+ in Politechnika Wroclawska, Poland
Relevant papers
Bayesian networks
Robustness
Uncertainty
Efficient Search for Relevance Explanations Using MAP-Independence in Bayesian Networks
We expand the idea of MAP-independence in Bayesian networks and explore its properties. We propose that this concept is related to explanation stability and that MAP-independence can be efficiently used to measure it and improve explanations in probabilistic graphical models.
Subgroup Discovery
Explainable Artificial Intelligence
Healthcare
Simple Explanations to Summarise Subgroup Discovery Outcomes: A Case Study Concerning Patient Phenotyping
The complexity of the subgroups created to characterise patients in phenotyping problems make the overall model difficult to understand. We propose a method with which to explain subgroup discovery, designed for the clinical context. We illustrate the suitability of the method in a clinical use case for an antimicrobial resistance problem and study the utility from a human-centric perspective.
AI in education
SHAP
Knowledge tracing
A SHAP-Inspired Method for Computing Interaction Contribution in Deep Knowledge Tracing
Deep knowledge tracing consists of predicting the probability of correctly answering a test or quiz question using the history of a particular learner’s previous question-answer interactions. In this work, an approach similar to Shapley Additive exPlanations (SHAP) to better understand DKT was used. The number of skills a learner must master to lead to improved learning outcomes in an explainable manner was first reduced.
Bayesian networks
Climate sciences
Open source
Interpreting Time-Varying Dynamic Bayesian Networks for Earth Climate Modelling
We introduce methods to explain how time-varying dynamic networks evolves qualitatively over time, and quantify these changes. In addition, we offer a functional open source library that streamlines the deployment of the model.
Get in Touch
If you have any doubt about my profile or just want to say hello, feel free to write me!