{ "id": "2210.11348", "version": "v1", "published": "2022-10-20T15:34:52.000Z", "updated": "2022-10-20T15:34:52.000Z", "title": "Hypernetworks in Meta-Reinforcement Learning", "authors": [ "Jacob Beck", "Matthew Thomas Jackson", "Risto Vuorio", "Shimon Whiteson" ], "comment": "Published at CoRL 2022", "categories": [ "cs.LG", "cs.AI", "cs.RO" ], "abstract": "Training a reinforcement learning (RL) agent on a real-world robotics task remains generally impractical due to sample inefficiency. Multi-task RL and meta-RL aim to improve sample efficiency by generalizing over a distribution of related tasks. However, doing so is difficult in practice: In multi-task RL, state of the art methods often fail to outperform a degenerate solution that simply learns each task separately. Hypernetworks are a promising path forward since they replicate the separate policies of the degenerate solution while also allowing for generalization across tasks, and are applicable to meta-RL. However, evidence from supervised learning suggests hypernetwork performance is highly sensitive to the initialization. In this paper, we 1) show that hypernetwork initialization is also a critical factor in meta-RL, and that naive initializations yield poor performance; 2) propose a novel hypernetwork initialization scheme that matches or exceeds the performance of a state-of-the-art approach proposed for supervised settings, as well as being simpler and more general; and 3) use this method to show that hypernetworks can improve performance in meta-RL by evaluating on multiple simulated robotics benchmarks.", "revisions": [ { "version": "v1", "updated": "2022-10-20T15:34:52.000Z" } ], "analyses": { "keywords": [ "meta-reinforcement learning", "degenerate solution", "novel hypernetwork initialization scheme", "real-world robotics task remains", "naive initializations yield poor performance" ], "note": { "typesetting": "TeX", "pages": 0, "language": "en", "license": "arXiv", "status": "editable" } } }