What augmentations are sensitive to hyper-parameters and why?

TL;DR

This paper investigates how different data augmentations in machine learning models vary in sensitivity to hyper-parameters, using interpretability methods to identify which augmentations are most affected and which are more stable.

Contribution

The study introduces a method to evaluate augmentation sensitivity to hyper-parameters and identifies which augmentations are more resilient or sensitive, enhancing robustness understanding.

Findings

Some augmentations are highly sensitive to hyper-parameters.

Certain augmentations show consistent performance regardless of hyper-parameter changes.

The approach helps in selecting more stable augmentations for robust models.

Abstract

We apply augmentations to our dataset to enhance the quality of our predictions and make our final models more resilient to noisy data and domain drifts. Yet the question remains, how are these augmentations going to perform with different hyper-parameters? In this study we evaluate the sensitivity of augmentations with regards to the model's hyper parameters along with their consistency and influence by performing a Local Surrogate (LIME) interpretation on the impact of hyper-parameters when different augmentations are applied to a machine learning model. We have utilized Linear regression coefficients for weighing each augmentation. Our research has proved that there are some augmentations which are highly sensitive to hyper-parameters and others which are more resilient and reliable.