Probeable DARTS with Application to Computational Pathology

TL;DR

This paper enhances neural architecture search for computational pathology by addressing DARTS hyperparameter tuning issues, resulting in more accurate and efficient models transferable across applications.

Contribution

It introduces a probing metric and adaptive optimization for DARTS, improving architecture search specifically for computational pathology tasks.

Findings

Searched networks outperform state-of-the-art in accuracy and efficiency

Enhanced transferability of architectures to new CPath applications

Improved robustness and prediction reliability

Abstract

AI technology has made remarkable achievements in computational pathology (CPath), especially with the help of deep neural networks. However, the network performance is highly related to architecture design, which commonly requires human experts with domain knowledge. In this paper, we combat this challenge with the recent advance in neural architecture search (NAS) to find an optimal network for CPath applications. In particular, we use differentiable architecture search (DARTS) for its efficiency. We first adopt a probing metric to show that the original DARTS lacks proper hyperparameter tuning on the CIFAR dataset, and how the generalization issue can be addressed using an adaptive optimization strategy. We then apply our searching framework on CPath applications by searching for the optimum network architecture on a histological tissue type dataset (ADP). Results show that the searched network outperforms state-of-the-art networks in terms of prediction accuracy and computation complexity. We further conduct extensive experiments to demonstrate the transferability of the searched network to new CPath applications, the robustness against downscaled inputs, as well as the reliability of predictions.