MIX'EM: Unsupervised Image Classification using a Mixture of Embeddings

07/18/2020
by   Ali Varamesh, et al.
14

We present MIX'EM, a novel solution for unsupervised image classification. Our model generates representations that by themselves are sufficient to drive a general-purpose clustering method to deliver high-quality classification without supervision. MIX'EM integrates an internal mixture of embeddings module into the contrastive visual representation learning framework to disentangle the representation space at the category level. It generates a set of embeddings from a visual representation and mixes them to construct the contrastive loss input. Parallel to the contrastive loss, we introduce three techniques to train MIX'EM and avoid a degenerate solution; (i) we maximize entropy across mixture components to diversify them, and (ii) minimize component entropy conditioned on instances to enforce a clustered embedding space. Applying (i) and (ii) lead to the emergence of semantic categories through the mixture coefficients, making it possible to (iii) apply an associative embedding loss to enforce semantic separability directly. Subsequently, we run K-means on the representations to acquire semantic classification, which outperforms the state-of-the-art by a large margin. We conduct extensive experiments and analyses on STL10, CIFAR10, and CIFAR100-20 datasets, achieving 78%, 82%, and 44% accuracy, respectively. Essential to robust high accuracy is using MIX'EM to initialize K-means. Finally, we report impressively high accuracy baselines (70% on STL10) achieved solely by applying K-means to the "normalized" representations learned using the contrastive loss.

READ FULL TEXT
research
05/11/2020

Prototypical Contrastive Learning of Unsupervised Representations

This paper presents Prototypical Contrastive Learning (PCL), an unsuperv...
research
09/14/2022

Joint Debiased Representation and Image Clustering Learning with Self-Supervision

Contrastive learning is among the most successful methods for visual rep...
research
09/26/2021

Cluster Analysis with Deep Embeddings and Contrastive Learning

Unsupervised disentangled representation learning is a long-standing pro...
research
03/30/2022

Fair Contrastive Learning for Facial Attribute Classification

Learning visual representation of high quality is essential for image cl...
research
11/23/2021

Exploring Non-Contrastive Representation Learning for Deep Clustering

Existing deep clustering methods rely on contrastive learning for repres...
research
05/05/2021

MiCE: Mixture of Contrastive Experts for Unsupervised Image Clustering

We present Mixture of Contrastive Experts (MiCE), a unified probabilisti...
research
04/23/2023

CoReFace: Sample-Guided Contrastive Regularization for Deep Face Recognition

The discriminability of feature representation is the key to open-set fa...

Please sign up or login with your details

Forgot password? Click here to reset