Enea Monzio Compagnoni

Enea Monzio Compagnoni

Research Intern at Google / Intrinsic and final-year PhD candidate at the University of Basel. Model optimization, cross-platform edge deployment, and stochastic optimization.

Talks and Presentations

2026

Understanding Optimizer Dynamics and Scaling Laws via SDEs

  1. Talk at Google DeepMind, 2026-08-11

On the Interaction of Batch Noise, Adaptivity, and Compression, under (L0, L1)-Smoothness: An SDE Approach

  1. Poster at International Conference on Machine Learning (ICML 2026), 2026-07-06–2026-07-11, Seoul, South Korea

Adaptive Methods Are Preferable in High Privacy Settings: An SDE Perspective

  1. Poster at International Conference on Learning Representations (ICLR 2026), 2026-04-23–2026-04-27, Rio de Janeiro, Brazil

Earlier talks and conference presentations

Unbiased and Sign Compression in Distributed Learning: Comparing Noise Resilience via SDEs

  1. Talk at Nokia Bell Labs, 2025-01-29, Stuttgart, Germany
  2. Talk at University of Milan, 2025-02-26, Milan, Italy
  3. Talk at Anima Anandkumar’s Lab, 2025-03-05, Caltech, Pasadena, USA
  4. Oral at International Conference on Artificial Intelligence and Statistics 2025, 2025-05-03, Mai Khao, Thailand
  5. Talk at Bocconi University, 2025-09-18, Milan, Italy

Adaptive Methods through the Lens of SDEs: Theoretical Insights on the Role of Noise

  1. Talk at EthonAI, 2024-05-29, Zürich, Switzerland
  2. Talk at University of Freiburg — Machine Learning Lab, 2025-01-16, Freiburg, Germany
  3. Poster at International Conference on Learning Representations 2025, 2025-04-24, Singapore

SDEs for Minimax Optimization

  1. Talk at Yahoo Research Series, 2024-03-12, online
  2. Poster at International Conference on Artificial Intelligence and Statistics 2024, 2024-05-04, Valencia, Spain

An SDE for Modeling SAM: Theory and Insights

  1. Talk at ELISE Research Workshop: Connections between Stochastic Calculus, Statistical Physics, and Optimization, 2023-06-22, University of Tübingen, Germany
  2. Talk at Yahoo Research Series, 2023-07-17, Munich, Germany
  3. Poster at International Conference on Machine Learning 2023, 2023-07-27, Honolulu, Hawaii, USA
  4. Poster at Scientific Advisory Boards der Universität Basel, 2023-11-27, Basel, Switzerland

On the Effectiveness of Randomized Signatures as Reservoir for Learning Rough Dynamics

  1. Talk at UBS Quant Research Series, 2022-09-30, Zürich, Switzerland
  2. Talk at Monte Verità Data Science Workshop 2022, 2022-10-03, Ticino, Switzerland
  3. Talk at University of Milan, 2023-02-10, Milan, Italy