Estimation of the implicit likelihoods of generative adversarial networks

Abstract

The disclosure discloses estimation of the implicit likelihoods of generative adversarial networks, and relates to the field of computer learning. The thriving of deep models and generative models provides approaches to model high dimensional distributions. Generative adversarial networks (GANs) can approximate data distributions and generate data samples from the learned data manifolds as well. Presented herein are embodiments to estimate the implicit likelihoods of GAN models. In one or more embodiments, a stable inverse function of the generator is learned with the help of a variance network of the generator. The local variance of the sample distribution may be approximated by the normalized distance in the latent space. Simulation studies and likelihood testing on data sets validate embodiments, which outperformed several baseline methods in these tasks. An embodiment was also applied to anomaly detection. Experiments show that the embodiments herein can achieve state-of-the-art anomaly detection performance.

Description

technical field [0001] The present disclosure generally relates to systems and methods for computer learning that can provide improved computer performance, features and uses. More specifically, the present disclosure relates to embodiments for estimating latent likelihoods of generative adversarial networks (GANs). Background technique [0002] Many real-world high-dimensional datasets cluster around low-dimensional unknown manifolds. Deep models provide new ways to estimate the density of extremely high-dimensional data. Generative models, such as generative adversarial networks (GANs), can also learn the distribution of high-dimensional datasets and generate samples. GANs typically use an adversarial loss as their training objective, which penalizes the discrepancy between the distribution of generated samples and real samples. Given an infinite approximate power, the original GAN ​​objective aims to minimize the Jensen-Shannon divergence between the real data distribu...

AI Technical Summary

Problems solved by technology

However, this method does not provide a qualitative way of estimating the likelihood of any given sample

Images