Edward Bergman

[Submitted on 4 Mar 2024 ( v1 ), last revised 18 Apr 2024 (this version, v2)] Title:Fast Benchmarking of Asynchronous Multi-Fidelity Optimization on Zero-Cost Benchmarks View a PDF of the paper titled Fast Benchmarking of Asynchronous Multi-Fidelity Optimization on Zero-Cost Benchmarks, by Shuhei Watanabe and Neeratyoy Mallik and Edward Bergman and Frank Hutter View PDF HTML (experimental) Abstract:While deep learning has celebrated many successes, its results often hinge on the meticulous...

Hyperparameters of Deep Learning (DL) pipelines are crucial for their downstream performance. While a large number of methods for Hyperparameter Optimization (HPO) have been developed, their incurred costs are often untenable for modern DL. Consequently, manual experimentation is still the most prevalent approach to optimize hyperparameters, relying on the researcher's intuition, domain knowledge, and cheap preliminary explorations.