Human-Understandable Decision Making for Visual Recognition

TL;DR

This paper introduces a framework that trains deep neural networks to align with human perception, making their decision-making process more interpretable while maintaining high accuracy in visual recognition tasks.

Contribution

The paper proposes a novel method that incorporates human perception priors into deep learning models to enhance interpretability without sacrificing performance.

Findings

The model provides interpretable explanations for predictions.

It maintains competitive recognition accuracy.

Effective on classical visual recognition tasks.

Abstract

The widespread use of deep neural networks has achieved substantial success in many tasks. However, there still exists a huge gap between the operating mechanism of deep learning models and human-understandable decision making, so that humans cannot fully trust the predictions made by these models. To date, little work has been done on how to align the behaviors of deep learning models with human perception in order to train a human-understandable model. To fill this gap, we propose a new framework to train a deep neural network by incorporating the prior of human perception into the model learning process. Our proposed model mimics the process of perceiving conceptual parts from images and assessing their relative contributions towards the final recognition. The effectiveness of our proposed model is evaluated on two classical visual recognition tasks. The experimental results and analysis confirm our model is able to provide interpretable explanations for its predictions, but also maintain competitive recognition accuracy.