TREND: Truncated Generalized Normal Density Estimation of Inception Embeddings for GAN Evaluation

TL;DR

TREND introduces a new GAN evaluation metric based on truncated generalized normal distribution of Inception embeddings, offering more accurate and robust assessments than traditional methods.

Contribution

The paper proposes TREND, a novel evaluation metric for GANs that models Inception features with a truncated generalized normal distribution for improved accuracy.

Findings

Reduces density estimation errors compared to existing metrics

Enhances robustness against sample size variations

Provides more reliable GAN evaluation results

Abstract

Evaluating image generation models such as generative adversarial networks (GANs) is a challenging problem. A common approach is to compare the distributions of the set of ground truth images and the set of generated test images. The Frech\'et Inception distance is one of the most widely used metrics for evaluation of GANs, which assumes that the features from a trained Inception model for a set of images follow a normal distribution. In this paper, we argue that this is an over-simplified assumption, which may lead to unreliable evaluation results, and more accurate density estimation can be achieved using a truncated generalized normal distribution. Based on this, we propose a novel metric for accurate evaluation of GANs, named TREND (TRuncated gEneralized Normal Density estimation of inception embeddings). We demonstrate that our approach significantly reduces errors of density estimation, which consequently eliminates the risk of faulty evaluation results. Furthermore, we show that the proposed metric significantly improves robustness of evaluation results against variation of the number of image samples.