Leonardo Cotta

Postdoc Fellow

💬 About

I’m a distinguished postdoc fellow at Vector Institute, where I’m hosted by Chris J. Maddison. I received my PhD degree in computer science at Purdue University advised by Bruno Ribeiro. Before that —in a distant and happy land— I was a BSc student (also in CS) at UFMG, Brazil. During my time as an undergrad I worked with distributed algorithms (at UFMG) and quantum computing theory (at University of Calgary).

Download my CV

📝 Research

I study and develop machine learning and causal inference methods for high-stakes applications of AI, such as science and recommendation engines. My long-term goal is to deliver new AI solutions that have real, tangible world impact. It’s going to be bitter, but now is our time to build. To this end, my work contributes algorithms and models that can leverage diverse, complex data to discover, evaluate, deliver, and monitor system interventions.

🗞️ Recent Pre-prints

  • Test-Time Fairness and Robustness in Large Language Models L Cotta, CJ Maddison arXiv, 2024 🌍 PAPER

  • Boosting the Predictive Power of Protein Representations with a Corpus of Text Annotations H Duan, M Skreta, L Cotta, EM Rajaonson, N Dhawan, A Aspuru-Guzik, CJ Maddison bioRxiv, 2024 🌍 PAPER

🎓 Selected Publications

For full list and metrics, see Google Scholar.

  • End-To-End Causal Effect Estimation from Unstructured Natural Language Data N Dhawan, L Cotta, K Ullrich, RG Krishnan, CJ Maddison NeurIPS, 2024 🌍 PAPER

  • Probabilistic Invariant Learning with Randomized Linear Classifiers L Cotta, G Yehuda, A Schuster, CJ Maddison NeurIPS, 2023 🌍 PAPER

  • Causal Lifting and Link Prediction L Cotta, B Bevilacqua, N Ahmed, B Ribeiro Proceedings of the Royal Society A, 2023 🌍 PAPER

  • Reconstruction for Powerful Graph Representations L Cotta, C Morris, B Ribeiro NeurIPS, 2021 🌍 PAPER

  • Unsupervised Joint k-node Graph Representations with Compositional Energy-Based Models L Cotta, CHC Teixeira, A Swami, B Ribeiro, 2020 NeurIPS, 2020 🌍 PAPER

  • Graph Pattern Mining and Learning through User-defined Relations CHC Teixeira, L Cotta, B Ribeiro, W Meira Jr ICDM, 2018 🌍 PAPER

  • AoT: Authentication and Access Control for the Entire IoT Device Life-Cycle AL Maia Neto, ALF Souza, Í Cunha, M Nogueira, IO Nunes, L Cotta, N Gentille, AAF Loureiro, DF Aranha, HK Patil, LB Oliveira ACM SenSys, 2016 🌍 PAPER

Recent Posts