BAHOP: Similarity-based Basin Hopping for A fast hyper-parameter search in WSI classification

TL;DR

This paper introduces BAHOP, a similarity-based basin hopping method that accelerates hyper-parameter search for WSI classification, significantly improving accuracy on out-of-domain data with faster inference.

Contribution

The paper presents BAHOP, a novel optimization approach that enables rapid domain-specific hyper-parameter tuning for WSI classification tasks.

Findings

BAHOP improves accuracy by 5% to 30% on out-of-domain WSIs.

BAHOP is approximately 5 times faster than traditional hyper-parameter search methods.

The method effectively enhances inference performance in domain adaptation scenarios.

Abstract

Pre-processing whole slide images (WSIs) can impact classification performance. Our study shows that using fixed hyper-parameters for pre-processing out-of-domain WSIs can significantly degrade performance. Therefore, it is critical to search domain-specific hyper-parameters during inference. However, searching for an optimal parameter set is time-consuming. To overcome this, we propose BAHOP, a novel Similarity-based Basin Hopping optimization for fast parameter tuning to enhance inference performance on out-of-domain data. The proposed BAHOP achieves 5\% to 30\% improvement in accuracy with $\times5$ times faster on average.