iiinova — where your digital self keeps evolving

Harley Wiltzer

@harwiltz.bsky.social

PhD student at Mila / McGill. Studying distributional RL for transfer across risk-sensitive utilities, and for long-horizon high-frequency decision-making.

Posts Replies Media

Reposted by Harley Wiltzer

Robin Ranjit Singh Chauhan @robinchauhan.bsky.social · 10/03/2025

E65: NeurIPS 2024 – Posters and Hallways 3 - Claire Bizon Monroc of Inria : WFCRL for Wind Farm Control Andrew Wagenmaker of @ucberkeleyofficial.bsky.social : Leveraging Simulation to Bridge Sim-to-Real Gap - @harwiltz.bsky.social of @mila-quebec.bsky.social : Multivariate Distributional RL (cont)

132

Harley Wiltzer @harwiltz.bsky.social · 09/12/2024

How can you 0-shot transfer predictions of long-term performance across reward functions *and* risk-sensitive utilities? We can do this via Distributional Successor Features. Our recent work introduces the 1st tractable & provably convergent algos for learning DSFs. #NeurIPS2024 #6704 12 Dec, 11-2

3164

Harley Wiltzer @harwiltz.bsky.social · 09/12/2024

In value-based RL, when decisions are made at high frequency, all hell breaks loose. Our paper "Action Gaps & Advantages in Continuous-Time Distributional RL" shows how Distributional RL sheds light on this, enabling high-frequency model-free risk-sensitive RL. #NeurIPS2024 #6410 13 Dec, 11-2

160

Reposted by Harley Wiltzer

Arthur Gretton @arthurgretton.bsky.social · 08/12/2024

Distributional SFs: enable 0-shot generalization of return *distribution* functions across a finite-dimensional reward function class "Foundations of Multivariate Distributional Reinforcement Learning" #NeurIPS2024 #6704 12 Dec 11am-2pm neurips.cc/virtual/2024... Wiltzer Farebrother Rowland

052