HyperQ-Opt: Q-learning for Hyperparameter Optimization

Md. Tarek Hasan · 2024

Read HyperQ-Opt: Q-learning for Hyperparameter Optimization by Md. Tarek Hasan (2024) 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-2412-17765.txt.

Description

Tuning a machine learning model's settings, called hyperparameters, can decide how well it performs, but searching the space of settings is slow. Tarek Hasan reframes that search as a sequence of decisions and applies Q-learning, a reinforcement learning method that learns from rewards. He reviews earlier work that cast tuning as a Markov decision process.

Those approaches refined settings step by step and found good or near-best setups within few trials, suggesting learning-based search can beat grid and random search. Hasan also flags gaps: most studies use discrete setting spaces and hand-made rules that may not scale. He argues policy-based tuning is a promising path toward faster, more scalable model building.

Catalog record

CREATORMd. Tarek Hasan
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE25 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-17765

Browse the full library · About the Tower of Babel