Summary of Verified Training For Counterfactual Explanation Robustness Under Data Shift, by Anna P. Meyer and Yuhao Zhang and Aws Albarghouthi and Loris D’antoni
Verified Training for Counterfactual Explanation Robustness under Data Shift
by Anna P. Meyer, Yuhao Zhang, Aws Albarghouthi, Loris D’Antoni
First submitted to arxiv on: 6 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary In this paper, the authors introduce VeriTraCER, an approach that jointly trains a classifier and an explainer to generate counterfactual explanations (CEs) that are robust to small changes in the underlying machine learning model. The generated CEs aim to describe what changes to an input are necessary to change its prediction to a desired class, while ensuring the explanations remain valid even if the model is updated periodically. This paper presents a new loss function that optimizes over both the classifier and explainer to ensure the verifiable robustness of the generated CEs to local model updates. The authors demonstrate the effectiveness of VeriTraCER by comparing its performance with state-of-the-art approaches in handling random initialization, leave-one-out, and distribution shifts. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary VeriTraCER is a new way for machine learning models to explain their decisions. It makes sure that these explanations stay true even if the model changes slightly over time. This is important because people want to understand why they were denied a loan or why their application was approved. The authors designed a special loss function that trains both the decision-making part of the model and its explanation-generating part together, so that they work well together. The results show that VeriTraCER performs better than other methods in keeping explanations accurate even when the model changes. |
Keywords
* Artificial intelligence * Loss function * Machine learning
