Tiago da Silva

Researcher @ MBZUAI

profile.jpg

I am a researcher working in Salem Lahlou’s group at MBZUAI. Prior to this, I obtained my PhD degree in Applied Mathematics from the School of Applied Mathematics of the Getulio Vargas Foundation (FGV) under the supervision of Diego Mesquita. I also hold a BSc and MSc degrees in Data Science from the same institution. During my PhD, I worked on (large-scale) amortized inference and statistical learning theory, with a special focus on GFlowNets.

In between my PhD and my current role, I worked on pricing and risk management projects as a data scientist and engineer.

Research-wise, I am broadly interested in approximate statistical inference and (Bayesian) learning theory.

selected publications

  1. ICLR
    When do GFlowNets learn the right Distribution?
    T. da Silva, R. Alves, E. SilvaA. SouzaV. GargS. Kaski, and D. Mesquita
    International Conference on Learning Representations, 2025
  2. ICLR
    Generalization and Distributed Learning of GFlowNets
    T. da SilvaA. SouzaO. RivasplataV. GargS. Kaski, and D. Mesquita
    International Conference on Learning Representations, 2025
  3. NeurIPS
    Streaming Bayes GFlowNets
    T. da SilvaD. de Souza, and D. Mesquita
    Conference on Neural Information Processing Systems, 2024
  4. NeurIPS
    On Divergence Measures for Training GFlowNets
    T. da SilvaE. Silva, and D. Mesquita
    Conference on Neural Information Processing Systems, 2024
  5. ICML
    Embarrassingly Parallel GFlowNets
    T. da SilvaA. SouzaL. CarvalhoS. Kaski, and D. Mesquita
    International Conference on Machine Learning, 2024
Thank you for visiting my website! Best way to reach me is via e-mail: tdsh97 [at] gmail [dot] com.