My thesis on computational intelligence (CI) was seminal to my current research on data science (DS), artificial intelligence (AI), and their impact on socially relevant problems.
I propose and collaborate with applied data science projects with both the public and private sectors. Partners include the Brazilian Judicial Branch and Ministry of Education, as well as regional and (multi-)national companies in fields as diverse as retail, telecommunications, and energy.
I supervise graduate students on theses involving deep and automated machine learning, as well as the intersection of multi-objective optimization with other CI domains, such as multi-dimensional visualization and dynamic optimization.
I have a deep concern for socially relevant problems, having for instance assisted in the fight against the COVID-19 pandemic through science publication and communication, in an attempt to counter the intensive disinformation campaign held in Brazil.
Currently, my main research focus is on promoting and sustaining accountability in artificial intelligence applications. My goal is to propose an end-to-end policy framework to be used by governments and AI developers alike to ensure that AI applications be devised and operated in an accountable way. To achieve this, I am applying for funding opportunities to structure an Observatory of the Potentially Incurred Social Damage from Artificial Intelligence (PISD.ai), which will survey current and future relevant real-world examples where the lack of appropriate AI regulation (potentially) incurs significant social damage. Critical examples include (i) the role of social media recommendation algorithms in disinformation and its impact on democracy, human rights, and public health, and; (ii) the disruptive nature of generative AI.
I was invited by the Benchmarking Network to attend the “BeMCO: Benchmarking in Multi-Criteria Optimization” workshop, which will take place from April 15th to 19th, 2024 at the Lorentz Center@Oort, Leiden, The Netherlands.
I was invited by Dr. Elizabeth Wanner to give a talk at Aston University, Birmingham, UK. The talk was hosted by the Aston Centre for Artificial Intelligence Research and Applications (ACAIRA) on October 16th, entitled “Promoting and sustaining accountability in artificial intelligence applications”.
Lecturer @ University of Stirling
Assistant professor @ Federal University of Rio Grande do Norte (UFRN)
Assistant professor @ Federal University of Paraíba (UFPB)
Alain Bensoussan fellowship @ European Research Consortium for Informatics and Mathematics (ERCIM)
Best paper award nomination @ Evolutionary Multi-Criterion Optimization (EMO)
Ph.D. degree in Engineering and Technology @ Université Libre de Bruxelles (ULB)
F.R.I.A doctoral fellowship @ Fonds de la Recherche Scientifique (FNRS)
Best paper award @ Brazilian Symposium on Augmented and Virtual Reality (SVR)
Financial details about the projects are provided in my CV.
Technological innovation cell @ Iberdrola Neoenergia COSERN
Applied research and human resource education in hardware technologies for artificial intelligence @ Huawei Telecommunications in Brazil
Generalization of metaheuristics for optimization problems with three or more objectives @ Fonds de la Recherche Scientifique (FNRS)
Information technology postgraduate apprenticeship (class of 2024) @ 5th Region Federal Regional Court (TRF5)
Information technology postgraduate apprenticeship (class of 2023) @ 5th Region Federal Regional Court (TRF5)
SmartMetropolis @ Multiple local and national government branches
Combinatorial optimization: metaheuristics and exact methods (COMEX) @ Belgian Federal Science Policy Office (BELSPO)
Algorithms for mobile robot path planning considering multiple objectives @ National Council for Scientific and Technological Development (CNPq)
Design configuration for the MMAS algorithm applied to the travelling salesman problem with dynamic demands @ Federal Center of Technological Education of Minas Gerais (CEFET-MG)
A case study on customer segmentation of a supermarket chain @ Federal University of Rio Grande do Norte (UFRN)
Sales forecasting for a supermarket chain in Natal, Brazil: an empirical assessment @ Federal University of Rio Grande do Norte (UFRN)
Assessing irace for automated machine and deep learning in computer vision @ Federal University of Rio Grande do Norte (UFRN)
Predspot: predicting crime hotspots with machine learning @ Federal University of Rio Grande do Norte (UFRN)
A metaheuristic approach to the high school timetabling problem at IFRN @ Federal University of Rio Grande do Norte (UFRN)
An exhaustive publication list with full author description is provided in my CV.
A computational study on ant colony optimization for the traveling salesman problem with dynamic demands
This paper was the main contribution from the first Ph.D. thesis I co-supervised, and demonstrates how multi-objective and dynamic optimization intersect. The relevance of this paper is evidenced by its best paper award nomination at the EMO 2019 conference, where a preliminary version of the journal paper was first published. In addition, this paper is a concrete example of how I bridge different research topics into multi-disciplinary work.
Comparing community mobility reduction between first and second COVID-19 waves
This paper was the main contribution of my efforts in science publication and communication to assist in the fight against the COVID-19 pandemic. Indeed, the first author of this paper is one of the undergraduate students that I helped mobilize in those initiatives. The relevance of this paper is evidenced by the number of different continents and COVID-19 waves included in the assessment. In addition, this paper is a concrete example of how I use computational intelligence in the context of socially relevant problems.
Automatically designing state-of-the-art multi-and many-objective evolutionary algorithms
A large-scale experimental evaluation of high-performing multi-and many-objective evolutionary algorithms
Automatic component-wise design of multiobjective evolutionary algorithms
These papers comprise the contributions of my Ph.D. thesis, having been accepted for publication prior to my defense or shortly after. Their relevance is evidenced by their ongoing impact on the evolutionary computation community, one of the most important in the context of CI, and by the rigorous journals where they were published. More importantly, these papers demonstrate how I am able to plan and deliver on a research project. In detail, each paper meets an specific objective of my thesis proposal, incrementally achieving the general objective of the project.
High school timetabling at a federal educational institute in Brazil
Retail sales forecasting for a Brazilian supermarket chain: an empirical assessment
Supermarket customer segmentation: a case study in a large Brazilian retail chain
Time-series features for predictive policing
Towards a crime hotspot detection framework for patrol planning
These papers comprise the contributions of data science M.Sc. theses I (co-)supervised in partnership with public and private institutions. The relevance of these papers is evidenced by the socially relevant scenarios they address. In detail, the first paper focuses on the Brazilian Federal Network of Vocational, Scientific and Technological Education, which provides education to over two million students, with over half of the students that declared income, gender, and ethnicity coming from low income families, being women, and self-declaring as non-white. The remainder 2022 papers use AI techniques to model different business processes in the 3rd largest retail supermarket chain in the Northeast of Brazil, and is instrumental to assess the impact of the COVID-19 pandemic in the industry. Finally, the 2018 papers address predictive policing to assist the local government in the forecasting of criminal occurrences.
Evaluating anytime performance on NAS-Bench-101
iSklearn: automated machine learning with irace
Comparing contextual embeddings for semantic textual similarity in Portuguese
These papers are the contributions of M.Sc. theses I supervised in deep and automated machine learning. The relevance of these papers is evidenced by the state-of-the-art techniques that were employed. In addition, the application domains considered are among the most relevant that use unstructured data, namely computer vision, natural language processing, and time series forecasting. Importantly, these papers demonstrate that I understand the technological complexity of current state-of-the-art AI models, their potential impact on society, and therefore their need for accountability.
Revisiting Pareto-optimal multi-and many-objective reference fronts for continuous optimization
Archiver effects on the performance of state-of-the-art multi-and many-objective evolutionary algorithms
An empirical assessment of the properties of inverted generational distance on multi-and many-objective optimization
These papers comprise follow-up works on my Ph.D. thesis. The relevance of these papers is evidenced by their ongoing impact on the evolutionary computation community, as well as the conferences where they were published, which are among the top-tier venues in their field. More importantly, these papers are a concrete example that the work I conducted in my Ph.D. was seminal to relevant future work. In addition, they demonstrate that I understand that seeking autonomy as an independent researcher does not mean discontinuing previous research.
During the climax of the first wave of the COVID-19 pandemic in 2020, I proposed and collaborated with several science communication efforts to help disseminate the guidelines from the World Health Organization (WHO) and counter the disinformation campaigns that were strongly affecting the social distance adherence in Brazil.
Promoting and sustaining accountability in artificial intelligence applications
@ Aston Centre for Artificial Intelligence Research and Applications (ACAIRA) - Aston University, Birmingham, UK
Brazilian Standard Classification of Education (CINE Brasil 2018)
@ Brazilian Ministry of Education - INEP
AutoML with Python - machine learning made easy(ish) @ Python Brasil 2018
A component-wise approach to multi-objective evolutionary algorithms: from flexible frameworks to automatic design
Automated algorithm engineering
pagmo C++ scientific library @ European Space Agency (ESA)
A practical introduction to irace
Automatic Generation of Multi-Objective ACO Algorithms for the Biobjective Knapsack
In addition to my formal collaborations through supervision and authoring, I have also met incredible people along my research career.
Check their whereabouts (updated eventually) at the interactive map at the beginning of this CV :blush: