These tools and metrics are designed to help AI actors develop and use trustworthy AI systems and applications that respect human rights and are fair, transparent, explainable, robust, secure and safe.
Fréchet Inception Distance (FID)
The Fréchet inception distance (FID) typically measures the quality of image generative models. More specifically, FID is a semimetric commonly applied to generative models based on generative adversarial networks (GANs), which was among the first generative modeling approaches to find success generating images.
FID can be loosely connected to Robustness, as a model that consistently produces low FID scores across various conditions may be considered more reliable and resilient in generating high-quality outputs. However, this connection is indirect, since FID does not explicitly measure robustness to adversarial conditions or failures, but rather the similarity of generated images to real ones.
About the metric
You can click on the links to see the associated metrics
Objective(s):
Purpose(s):
Lifecycle stage(s):
Target users:
