cv

This is a summary of my CV. For a complete version, please download the PDF.

Personal Information

Full name Lucas Emanuel Resck Domingues
Position MSc student in Mathematical Modeling
School Fundação Getulio Vargas
Place Rio de Janeiro, Brazil
Email lucas.resck@fgv.br
Research interests Machine Learning (ML), Natural Language Processing (NLP), and Explainable Artificial Intelligence (XAI). I am particularly interested in improving the degree of explainability of ML and NLP models.

Education

  • 2022 - present
    MSc, Mathematical Modeling
    Fundação Getulio Vargas
  • 2018 - 2021
    BSc, Applied Mathematics
    Fundação Getulio Vargas
    • GPA: 3.86/4.0 --- 1st in class, 9.66/10.0, lowest passing grade of 6

Research Experience

  • 2020 - present
    Master's and Undergraduate Researcher
    Visual Data Science Lab
    • Fundação Getulio Vargas
    • Supervisor: Jorge Poco
    • Design of a novel explainer for GNN node classification
      • "Distill n' Explain" (AISTATS 2023) first distills the original GNN into an interpretable one and then explains the latter.
      • Designed and proved lemmas and theorems that guarantee the method's explanation faithfulness.
      • The proposed explainer outperformed previous methods in explanation accuracy while being orders of magnitude faster.
    • Development of a visual analytics system to explore citations in legal documents
      • "LegalVis" (TVCG 2023, VIS 2022, BSc Thesis) employs ML, NLP, XAI, and data visualization to infer non-explicit citations in Brazilian legal documents.
      • Tested a diverse set of NLP classifiers (Transformers, word embeddings, and bag-of-words) and achieved high accuracy (96%) in identifying citations.
      • Employed a model-agnostic explainer (LIME) to explain the inferred citations.
  • 2022
    Visiting Scholar (3 months)
    Vision, Language, and Learning Lab
    • Rice University, Houston, USA
    • Supervisor: Vicente Ordóñez
    • Explored training data attribution methods, e.g., influence functions, and ways to improve them. This activity continued after the visit.