ELLIS Unit Milan - RIKEN AIP Joint Workshop

RIKEN AIP and ELLIS Unit Milan run a two-day online joint workshop on machine learning theory and applications. Day 1 (May 7, 21:00–00:40 JST) opens with introductions from Nicolò Cesa-Bianchi and Masashi Sugiyama, then talks on reward modeling in reinforcement learning, dynamical analysis of deep networks and transformers, privacy-aware Bayesian networks, adaptive learning rates, implicit bias of gradient descent, SGD on k-juntas, and causal additive models. Day 2 (May 8, 16:00–23:30 JST) covers Laplacian representations for RL, regret minimization for piecewise linear rewards, adaptation costs in online learning, pure exploration, UMAP evaluation, task arithmetic for language models, online learning and game theory, target sample complexity, unconstrained linear bandits, AI-driven RIS-SWIPT for IoT, and time-frequency learning for time series forecasting. The program is research-grade and aimed at ML researchers and graduate students. Speakers are split between the two institutes, and sessions run live in English over the two evenings JST.