Scholarlink

This chapter argues the importance of robustness, a key aspect for the success of explainable AI. The chapter introduces the robustness problem for two key types of explanations: counterfactual explanations, which suggest minimal changes to inputs that would alter the prediction, and feature attribution explanations, which identify the most influential features in an input that are responsible for a classification outcome.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Introduction

  • 章节

  • 发表日期：

    2025年

查看全文
  • Francesco Leofante,
  • Matthew Wicker

摘要

This chapter argues the importance of robustness, a key aspect for the success of explainable AI. The chapter introduces the robustness problem for two key types of explanations: counterfactual explanations, which suggest minimal changes to inputs that would alter the prediction, and feature attribution explanations, which identify the most influential features in an input that are responsible for a classification outcome.