Meta-Reinforcement Learning with Discrete World Models for Adaptive Load Balancing DOI: 10.1145/YYYYYYY.YYYYYYY Conference: 2025 ACM Southeast Conference; April 24–26, 2025; Cape Girardeau, MO, USA 2025 ACM Southeast Conference (ACMSE 2025), April 24–26, 2025, Cape Girardeau, MO, USA Price: 15.00 IS
Cameron Redovian · 2025
Read Meta-Reinforcement Learning with Discrete World Models for Adaptive Load Balancing DOI: 10.1145/YYYYYYY.YYYYYYY Conference: 2025 ACM Southeast Conference; April 24–26, 2025; Cape Girardeau, MO, USA 2025 ACM Southeast Conference (ACMSE 2025), April 24–26, 2025, Cape Girardeau, MO, USA Price: 15.00 IS by Cameron Redovian (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-08872.txt.
Description
Cameron Redovian presents a learning system for balancing work across computers in this 2025 study. Operating systems must hand tasks to processors as loads shift by the second, and a poor balance wastes power and slows jobs. Redovian joins a meta-learning method, which learns how to adapt fast, with a world-model agent that dreams up outcomes before acting.
In trials, the combined system adapted to new workloads with little retraining. It beat a standard actor-critic method in both fixed and changing tests. It also resisted catastrophic forgetting, the failure where new learning wipes old skills, holding its level as workload sizes and shapes varied. The author argues the design suits real system management, where conditions rarely stay still.
Catalog record
| CREATOR | Cameron Redovian |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 28 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-08872