A Brief History of Fréchet Distances: From Curves and Probability Laws to FID

2026-04-23

AI Summary

This study systematically traces the century-long evolution of the Fréchet distance—from its original 1906 formulation in abstract metric spaces and its 1957 reinterpretation via couplings of probability measures to its 1990s adaptation as a curve-matching algorithm—thereby establishing, for the first time, a unified theoretical bridge between its geometric and probabilistic interpretations. Through historical document analysis and optimal transport theory, the work demonstrates that the Fréchet Inception Distance (FID), widely used to evaluate generative models, is in fact a special case of the Wasserstein-2 distance operating in the feature space induced by deep neural networks. Beyond clarifying the historical and conceptual trajectory of the Fréchet distance, this research provides an English translation appendix of key original texts and offers a deeper theoretical grounding for understanding FID as a metric in modern machine learning.

Technology Category

• Computer Vision: Other Foundations of Computer Vision
• Machine Learning: Other Foundations of Machine Learning
• Cognitive Modeling & Cognitive Systems: Other Foundations of Cognitive Modeling & Systems

Application Category

• Web Mining and Content Analysis: Models for Web evolution
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs

Abstract

This note provides a chronological account of Fréchet distances, starting with Maurice Fréchet's 1906 doctoral thesis on distances in abstract sets and tracing the Fréchet distance between polygonal curves and its algorithmic computation in the 1990s. It then continues with his 1957 paper on a coupling-based distance between probability laws with a brief glimpse of Wasserstein distance and optimal transport. We further attempt to draw connections between the distributional, coupling-based facet of Fréchet distances on probability laws and the geometric facet on curves. The note ends with a modern use case, the Fréchet Inception Distance (FID) in the era of deep generative model evaluation, interpretable as the Wasserstein-2 distance between multivariate Gaussians in a learned feature space. An appendix includes \TeX{}ified faithful English translations of Fréchet's 1906 thesis and 1957 paper, and Lévy's 1950 note for reader convenience.

Problem

• Fréchet distance
• probability laws
• Wasserstein distance
• optimal transport
• generative models

Innovation

• Fréchet distance
• Wasserstein distance
• optimal transport
• Fréchet Inception Distance
• probability laws