search

christophM / interpretable-ml-book

Book about interpretable machine learning
https://christophm.github.io/interpretable-ml-book/
Other
4.76k stars 1.06k forks source link

Disagreement of Explainers #347

Open vinitra opened 2 years ago

vinitra commented 2 years ago

Hi Christoph -- great work with the recent additions to the book; I really liked the neural net chapters. I recommend this book to everyone. :)

I might have missed it, but I think there's a great space to talk about recent studies in disagreement for black box explainers at the end of Chapter 9. I had a recent publication on this in the XAI for education space. Around the same time, there was a great ArXiv preprint by a team from Harvard / MIT / CMU and Drexel that discusses the more general implications on different popular datasets, also with a data scientist user study.

Of course, this 2019 Nature article asking us to use interpretable models instead of explainability methods from Cynthia Rudin is a predecessor and a classic.

Thanks, Vinitra

christophM commented 2 years ago

Hi Vinitra

Thanks a lot for sharing your paper(s), and also thank you for your nice words. I will take a look.

Best, Christoph

Am Mi., 21. Sept. 2022 um 13:20 Uhr schrieb Vinitra Swamy < @.***>:

Hi Christoph -- great work with the recent additions to the book; I really liked the neural net chapters. As I wrote you a few years ago, I recommend this book to any interested students working in trustworthy AI. :)

I might have missed it, but I think there's a great space to talk about recent studies in disagreement for black box explainers at the end of Chapter 9. I had a recent publication on this in the XAI for education space https://educationaldatamining.org/edm2022/proceedings/2022.EDM-long-papers.9/2022.EDM-long-papers.9.pdf. Around the same time, there was a great ArXiv preprint by a team from Harvard / MIT / CMU and Drexel https://arxiv.org/abs/2202.01602 that discusses the more general implications on different popular datasets, also with a data scientist user study.

Of course, this 2019 Nature article asking us to use interpretable models instead of explainability methods from Cynthia Rudin https://www.nature.com/articles/s42256-019-0048-x is a predecessor and a classic.

Thanks, Vinitra

— Reply to this email directly, view it on GitHub https://github.com/christophM/interpretable-ml-book/issues/347, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAMOOZBVVQ3JK252FRCQUGDV7LVPLANCNFSM6AAAAAAQR632MQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>

-- Christoph Molnar https://christophmolnar.com/

Machine Learning Expert | Author of Interpretable Machine Learning https://bit.ly/3K3AV1y and Modeling Mindsets https://leanpub.com/modeling-mindsets