CHIRPs: Change-Induced Regret Proxy metrics for Lifelong Reinforcement Learning

no code implementations • 5 Sep 2024 • John Birkbeck, Adam Sobey, Federico Cerutti, Katherine Heseltine Hurley Flynn, Timothy J. Norman

We demonstrate two uses for these metrics: for learning, an agent that clusters MDPs based on a CHIRP metric achieves $17\%$ higher average returns than three existing agents in a sequence of MetaWorld tasks.

reinforcement-learning Reinforcement Learning

From Intelligent Agents to Trustworthy Human-Centred Multiagent Systems

no code implementations • 5 Oct 2022 • Mohammad Divband Soorati, Enrico H. Gerding, Enrico Marchioni, Pavel Naumov, Timothy J. Norman, Sarvapali D. Ramchurn, Bahar Rastegari, Adam Sobey, Sebastian Stein, Danesh Tarpore, Vahid Yazdanpanah, Jie Zhang

The Agents, Interaction and Complexity research group at the University of Southampton has a long track record of research in multiagent systems (MAS).

Automation for Interpretable Machine Learning Through a Comparison of Loss Functions to Regularisers

no code implementations • NeurIPS 2021 • Amy Parkes, Adam Sobey, Josef Camilleri, Dominic Hudson

This paper explores the use of the Fit to Median Error measure in machine learning regression automation, using evolutionary computation in order to improve the approximation of the ground truth.

BIG-bench Machine Learning Interpretable Machine Learning +1