Example Perplexity

TL;DR

This paper introduces a method to measure the difficulty of classifying individual examples for deep neural networks, aiming to understand factors influencing classification perplexity.

Contribution

It proposes a novel approach to quantify example perplexity and analyzes factors affecting classification difficulty in DNNs.

Findings

Method effectively measures example perplexity

Factors influencing high perplexity are identified

Resources available at provided GitHub link

Abstract

Some examples are easier for humans to classify than others. The same should be true for deep neural networks (DNNs). We use the term example perplexity to refer to the level of difficulty of classifying an example. In this paper, we propose a method to measure the perplexity of an example and investigate what factors contribute to high example perplexity. The related codes and resources are available at https://github.com/vaynexie/Example-Perplexity.