Summary of Gnothi Seauton: Empowering Faithful Self-interpretability in Black-box Transformers, by Shaobo Wang et al.
Gnothi Seauton: Empowering Faithful Self-Interpretability in Black-Box Transformers
by Shaobo Wang, Hongxuan Tang, Mingyang Wang, Hongrui Zhang, Xuyang Liu, Weiya Li, Xuming Hu, Linfeng Zhang
First submitted to arxiv on: 29 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Computer Science and Game Theory (cs.GT)
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 paper addresses the Explainable AI (XAI) debate between self-interpretable models and post-hoc explanations for black-box models. Self-interpretable models, such as concept-based networks, provide insights by connecting decisions to human-understandable concepts but often struggle with performance and scalability. Post-hoc methods like Shapley values are theoretically robust but computationally expensive and resource-intensive. To bridge this gap, the authors propose AutoGnothi, a novel method that combines strengths from both approaches. This pipeline integrates a small side network into the black-box model to generate Shapley value explanations without changing original network parameters. AutoGnothi reduces memory, training, and inference costs while outperforming traditional parameter-efficient methods. The authors demonstrate accurate explanations for vision and language tasks with minimal overhead using extensive experiments. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about making AI models more understandable. Currently, there are two ways to do this: either the model can understand itself or we can add explanations later. Both approaches have their own problems. The first one is not very good at predicting things, while the second one takes a lot of time and computer power. To solve this problem, the authors created a new method called AutoGnothi. It combines the strengths of both approaches to make AI models that can predict and explain what they’re doing with minimal extra effort. This is important because we need AI models that are not only good at predicting things but also understandable by humans. |
Keywords
» Artificial intelligence » Inference » Parameter efficient
