Explaining Model Overfitting in CNNs via GMM Clustering

TL;DR

This paper introduces a GMM-based clustering method to analyze CNN filters, providing insights into overfitting by identifying anomaly filters and offering a new perspective on model evaluation.

Contribution

It proposes a universal GMM clustering approach to assess CNN filters and analyze their relation to overfitting, applicable across various architectures without modifications.

Findings

GMM clustering effectively identifies anomaly filters.

Analysis reveals links between anomaly filters and overfitting.

Method applicable to models like AlexNet and LeNet-5.

Abstract

Convolutional Neural Networks (CNNs) have demonstrated remarkable prowess in the field of computer vision. However, their opaque decision-making processes pose significant challenges for practical applications. In this study, we provide quantitative metrics for assessing CNN filters by clustering the feature maps corresponding to individual filters in the model via Gaussian Mixture Model (GMM). By analyzing the clustering results, we screen out some anomaly filters associated with outlier samples. We further analyze the relationship between the anomaly filters and model overfitting, proposing three hypotheses. This method is universally applicable across diverse CNN architectures without modifications, as evidenced by its successful application to models like AlexNet and LeNet-5. We present three meticulously designed experiments demonstrating our hypotheses from the perspectives of model behavior, dataset characteristics, and filter impacts. Through this work, we offer a novel perspective for evaluating the CNN performance and gain new insights into the operational behavior of model overfitting.