Summary of Context-aware Feature Attribution Through Argumentation, by Jinfeng Zhong et al.
Context-aware feature attribution through argumentation
by Jinfeng Zhong, Elsa Negre
First submitted to arxiv on: 24 Oct 2023
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Applications (stat.AP)
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 This research paper proposes a novel framework for feature attribution called Context-Aware Feature Attribution Through Argumentation (CA-FATA). The authors aim to improve the current state-of-the-art by developing an interpretable and context-aware method that can accurately attribute features in AI systems. CA-FATA harnesses the power of argumentation, treating each feature as an argument that can support, attack, or neutralize a prediction. This framework formulates feature attribution as an argumentation procedure, making it inherently interpretable. Additionally, CA-FATA integrates side information such as users’ contexts, leading to more accurate predictions. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Feature attribution is important in machine learning and data analysis because it helps identify the most important features for predicting outcomes. Current methods have limitations, including lower accuracy with General Additive Models (GAMs), difficulty interpreting gradient-based methods, and stability and fidelity issues with surrogate models. The new CA-FATA framework aims to address these limitations by using argumentation to attribute features and incorporating users’ contexts. |
Keywords
» Artificial intelligence » Machine learning
