Summary of Tabularbench: Benchmarking Adversarial Robustness For Tabular Deep Learning in Real-world Use-cases, by Thibault Simonetto et al.
TabularBench: Benchmarking Adversarial Robustness for Tabular Deep Learning in Real-world Use-cases
by Thibault Simonetto, Salah Ghamizi, Maxime Cordy
First submitted to arxiv on: 14 Aug 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 This research paper proposes a comprehensive benchmark for evaluating the adversarial robustness of tabular deep learning classification models. The authors develop TabularBench, a standardized benchmark that assesses the performance of 200 models across five critical scenarios in finance, healthcare, and security. They implement seven robustification mechanisms inspired by state-of-the-art defenses in computer vision and train and assess their models with real datasets and hundreds of thousands of synthetic inputs. The authors open-source their library, providing API access to pre-trained robust tabular models and the largest datasets of real and synthetic tabular inputs. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Tabular deep learning is a type of artificial intelligence that’s used for tasks like predicting patient outcomes or identifying financial fraud. But just like how humans can’t always rely on each other, these AI systems aren’t perfect either. Sometimes they get tricked into making wrong predictions by fake data. This paper tries to fix this problem by creating a set of tests, called TabularBench, that helps developers see how well their models can handle tricky data. The authors also come up with new ways to make the models more robust and share them online so others can use. |
Keywords
* Artificial intelligence * Classification * Deep learning
