arXiv Open Access 2019

Variational Information Bottleneck for Unsupervised Clustering: Deep Gaussian Mixture Embedding

Yigit Ugur George Arvanitakis Abdellatif Zaidi

Lihat Sumber

Abstrak

In this paper, we develop an unsupervised generative clustering framework that combines the Variational Information Bottleneck and the Gaussian Mixture Model. Specifically, in our approach, we use the Variational Information Bottleneck method and model the latent space as a mixture of Gaussians. We derive a bound on the cost function of our model that generalizes the Evidence Lower Bound (ELBO) and provide a variational inference type algorithm that allows computing it. In the algorithm, the coders' mappings are parametrized using neural networks, and the bound is approximated by Monte Carlo sampling and optimized with stochastic gradient descent. Numerical results on real datasets are provided to support the efficiency of our method.

Topik & Kata Kunci

cs.LG cs.IT stat.ML

Penulis (3)

Yigit Ugur

George Arvanitakis

Abdellatif Zaidi

Format Sitasi

APA MLA BibTeX

Ugur, Y., Arvanitakis, G., Zaidi, A. (2019). Variational Information Bottleneck for Unsupervised Clustering: Deep Gaussian Mixture Embedding. https://arxiv.org/abs/1905.11741

Akses Cepat

Lihat di Sumber

Informasi Jurnal

Tahun Terbit: 2019
Bahasa: en
Sumber Database: arXiv
Akses: Open Access ✓