https://github.com/wangyongjie-ntu/reference-papers

https://github.com/wangyongjie-ntu/reference-papers

Last synced: 3 months ago
JSON representation

• Host: GitHub
• URL: https://github.com/wangyongjie-ntu/reference-papers
• Owner: wangyongjie-ntu
• Created: 2020-06-10T08:06:08.000Z (almost 5 years ago)
• Default Branch: master
• Last Pushed: 2020-09-24T13:56:18.000Z (over 4 years ago)
• Last Synced: 2025-01-09T19:57:39.818Z (4 months ago)
• Size: 6.84 KB
• Stars: 0
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Reference-Papers

## Explainable AI

Survey paper: [A survey of methods for explaining black box models](http://arxiv.org/abs/1802.01933), ACM Computing Surveys, 2018

Blur integrated gradient: [Attribution in Scale and Space](https://arxiv.org/pdf/2004.03383.pdf), CVPR 2020 

DeepLIFT: [Learning important features through propagating activation differences](http://arxiv.org/abs/1704.02685), ICML 2017

Integrated Gradients: [Axiomatic attribution for deep networks](http://arxiv.org/abs/1703.01365), ICML 2017

LRP, [On pixel-wise explanations for non-linear classifier decisions by layer-wise relevance propagation](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4498753/), PloS one 2015

more papers: [https://github.com/wangyongjie-ntu/Awesome-explainable-AI](https://github.com/wangyongjie-ntu/Awesome-explainable-AI)

## Actionable Recourse (Counterfactual Explanation)

This task mainly targets to find an explanation of a given instance, such that the explanation can change the prediction(Ususally from an undesirable outcome to an ideal one). The explanation itself is also a valid instance, or changing of current instance(can reduce to another instance implicitly) in the feature space. This task simply extracts knowledge from the black-box models. This task belongs to the instance-level explanation. Quite interesting to dive deeper!!!

[Algorithmic recourse under imperfect causal knowledge a probabilistic approach](https://arxiv.org/abs/2006.06831), Arxiv preprint 2020

[Algorithmic Recourse: from Counterfactual Explanations to Interventions](https://arxiv.org/abs/2002.06278), Arxiv preprint 2020

[Learning Model-Agnostic Counterfactual Explanations for Tabular Data](https://dl.acm.org/doi/abs/10.1145/3366423.3380087),  ACM WWW 2020

[Explaining Machine Learning Classifiers through Diverse Counterfactual Explanations](https://arxiv.org/pdf/1905.07697.pdf), ACM FAT 2020

[Towards Realistic Individual Recourse and Actionable Explanations in Black-Box Decision Making Systems](https://arxiv.org/pdf/1907.09615.pdf), Arxiv preprint 2019

[Actionable Recourse in Linear Classification](https://dl.acm.org/doi/pdf/10.1145/3287560.3287566), FAT 2019

[EQUALIZING RECOURSE ACROSS GROUPS](https://arxiv.org/abs/1909.03166),  Arxiv preprint 2019

[Efficient Search for Diverse Coherent Explanations](https://arxiv.org/pdf/1901.04909.pdf), ACM FAT 2019

[Explaining image classifiers by counterfactual generation](https://arxiv.org/pdf/1807.08024.pdf), ICLR 2019

[Counterfactual Explanations without Opening the Black Box: Automated Decisions and the GDPR](https://arxiv.org/abs/1711.00399), Hardvard Journal of Law & Technology 2018 (strong recommend)

[Inverse Classification for Comparison-based Interpretability in Machine Learning](https://arxiv.org/pdf/1712.08443.pdf), Arxiv 2017