Deep clustering using adversarial net based clustering loss

TL;DR

This paper introduces an adversarial net-based deep clustering method that reformulates the loss function to improve clustering performance, leveraging adversarial training to approximate divergence measures.

Contribution

It presents a novel adversarial approach to deep clustering that replaces traditional closed-form loss functions with a discriminator-based framework, enabling better divergence approximation.

Findings

Achieved comparable or superior results on MNIST, REUTERS10K, and CIFAR10 datasets.

Demonstrated the effectiveness of the adversarial clustering loss in deep clustering tasks.

Abstract

Deep clustering is a recent deep learning technique which combines deep learning with traditional unsupervised clustering. At the heart of deep clustering is a loss function which penalizes samples for being an outlier from their ground truth cluster centers in the latent space. The probabilistic variant of deep clustering reformulates the loss using KL divergence. Often, the main constraint of deep clustering is the necessity of a closed form loss function to make backpropagation tractable. Inspired by deep clustering and adversarial net, we reformulate deep clustering as an adversarial net over traditional closed form KL divergence. Training deep clustering becomes a task of minimizing the encoder and maximizing the discriminator. At optimality, this method theoretically approaches the JS divergence between the distribution assumption of the encoder and the discriminator. We demonstrated the performance of our proposed method on several well cited datasets such as MNIST, REUTERS10K and CIFAR10, achieving on-par or better performance with some of the state-of-the-art deep clustering methods.