https://github.com/andreartelt/analyzinginfluencetrainingsamplesexplanations

"Analyzing the Influence of Training Samples on Explanations" by André Artelt et al.

https://github.com/andreartelt/analyzinginfluencetrainingsamplesexplanations

Science Score: 23.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • CITATION.cff file
  • codemeta.json file
    Found codemeta.json file
  • .zenodo.json file
  • DOI references
  • Academic publication links
    Links to: arxiv.org
  • Academic email domains
  • Institutional organization owner
  • JOSS paper metadata
  • Scientific vocabulary similarity
    Low similarity (8.7%) to scientific vocabulary
Last synced: 10 months ago · JSON representation

Repository

"Analyzing the Influence of Training Samples on Explanations" by André Artelt et al.

Basic Info
  • Host: GitHub
  • Owner: andreArtelt
  • License: mit
  • Language: Python
  • Default Branch: main
  • Homepage:
  • Size: 905 KB
Statistics
  • Stars: 1
  • Watchers: 1
  • Forks: 1
  • Open Issues: 0
  • Releases: 0
Created almost 2 years ago · Last pushed about 1 year ago
Metadata Files
Readme License

README.md

Analyzing the Influence of Training Samples on Explanations

This repository contains the implementation of the experiments as proposed in the paper Towards Understanding the Influence of Training Samples on Explanations by André Artelt and Barbara Hammer.

Abstract

Explainable AI (XAI) is widely used to analyze AI systems' decision-making, such as providing counterfactual explanations for recourse. When unexpected explanations occur, users may want to understand the training data properties shaping them. Under the umbrella of data valuation, first approaches have been proposed that estimate the influence of data samples on a given model. This process not only helps determine the data's value, but also offers insights into how individual, potentially noisy, or misleading examples affect a model, which is crucial for interpretable AI. In this work, we apply the concept of data valuation to the significant area of model evaluations, focusing on how individual training samples impact a model's internal reasoning rather than the predictive performance only. Hence, we introduce the novel problem of identifying training samples shaping a given explanation or related quantity, and investigate the particular case of the cost of computational recourse. We propose an algorithm to identify such influential samples and conduct extensive empirical evaluations in two case studies.

Experiments

All experiments are implemented in Python and were tested with Python 3.8 -- required dependencies are listed in REQUIREMENTS.txt. The experiments themself are implemented in experiments.py and eval_experiments.py.

The proposed algorithm (Algorithm 1 in the paper) is implemented in gradientdatashapley.py.

License

MIT license - See LICENSE.

How to cite

The version in this repository constitutes an extended version of the IJCAI 2024 XAI workshop version.

You can cite either the workshop version or this version on arXiv.

Owner

  • Name: André Artelt
  • Login: andreArtelt
  • Kind: user
  • Location: Germany
  • Company: Bielefeld University

PhD student

GitHub Events

Total
  • Watch event: 1
  • Push event: 1
  • Public event: 1
  • Fork event: 1
Last Year
  • Watch event: 1
  • Push event: 1
  • Public event: 1
  • Fork event: 1