Performance of Gaussian Mixture Model Classifiers on Embedded Feature Spaces

TL;DR

This paper evaluates Gaussian Mixture Model classifiers on embedded features from CLIP and ImageBind, proposing a low-parameter GMM classifier and finding that a single Gaussian often suffices for class separation due to the nature of the embedded spaces.

Contribution

It introduces a low-parameter GMM classifier and assesses its performance on CLIP and ImageBind embeddings, highlighting the effectiveness of single Gaussian components per class.

Findings

Single Gaussian components often suffice for class separation.

ImageBind generally outperforms CLIP on image classification tasks.

Embedded spaces are naturally concentrated due to contrastive training.

Abstract

Data embeddings with CLIP and ImageBind provide powerful features for the analysis of multimedia and/or multimodal data. We assess their performance here for classification using a Gaussian Mixture models (GMMs) based layer as an alternative to the standard Softmax layer. GMMs based classifiers have recently been shown to have interesting performances as part of deep learning pipelines trained end-to-end. Our first contribution is to investigate GMM based classification performance taking advantage of the embedded spaces CLIP and ImageBind. Our second contribution is in proposing our own GMM based classifier with a lower parameters count than previously proposed. Our findings are, that in most cases, on these tested embedded spaces, one gaussian component in the GMMs is often enough for capturing each class, and we hypothesize that this may be due to the contrastive loss used for training these embedded spaces that naturally concentrates features together for each class. We also observed that ImageBind often provides better performance than CLIP for classification of image datasets even when these embedded spaces are compressed using PCA.