Neural clustering based on implicit maximum likelihood

Abstract Clustering is one of the most fundamental unsupervised learning tasks with numerous applications in various fields. methods based on neural networks, called deep clustering methods, leverage representational power networks to enhance performance. ClusterGan constitutes a generative method that exploits adversarial (GANs) perform clustering. However, it inherits some deficiencies GANs, such as mode collapse, vanishing gradients and training instability. In order tackle those deficiencies, approach implicit maximum likelihood estimation (IMLE) has been recently proposed. this paper, we present clustering, which adopts ideas from both ClusterGAN IMLE. The proposed compared other synthetic real datasets, demonstrating promising results.