Summary of Shapley Marginal Surplus For Strong Models, by Daniel De Marchi et al.
Shapley Marginal Surplus for Strong Models
by Daniel de Marchi, Michael Kosorok, Scott de Marchi
First submitted to arxiv on: 16 Aug 2024
Categories
- Main: Machine Learning (stat.ML)
- Secondary: Machine Learning (cs.LG)
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 Shapley values have been widely used in machine learning to explain model predictions and estimate the importance of covariates. While model-based Shapley values are accurate explainers of model predictions, machine learning models themselves often fail to accurately explain the underlying data-generating process (DGP), even when highly predictive. This is particularly true for interrelated or noisy variables, where a highly predictive model may not account for these relationships. As a result, explanations of a trained model’s behavior may not provide meaningful insight into the DGP. To address this issue, we introduce Shapley Marginal Surplus for Strong Models (SMS4), which samples the space of possible models to estimate feature importance. SMS4 outperforms other popular feature importance methods in inferential capabilities. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Machine learning models are used to predict and explain data, but they don’t always tell us how they work or what’s important. Shapley values help us understand model predictions, but even accurate models can fail to show us what’s going on with the real-world data that created them. A new method called SMS4 (Shapley Marginal Surplus for Strong Models) tries to fix this problem by looking at many possible models and figuring out which features are most important. SMS4 beats other popular methods in showing us what really matters. |
Keywords
» Artificial intelligence » Machine learning
