Summary of Optimizing Calibration by Gaining Aware Of Prediction Correctness, By Yuchi Liu et al.
Optimizing Calibration by Gaining Aware of Prediction Correctness
by Yuchi Liu, Lei Wang, Yuli Zou, James Zou, Liang Zheng
First submitted to arxiv on: 19 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG); Machine Learning (stat.ML)
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 The proposed post-hoc calibration objective is derived from the goal of model calibration, which aims to align confidence with prediction correctness. The Cross-Entropy (CE) loss is widely used for calibrator training, but it has intrinsic limitations, such as increasing confidence on misclassified samples. To address this issue, a new calibration objective function is introduced that decreases model confidence on wrongly predicted samples and increases confidence on correctly predicted ones. This approach uses transformed versions of the sample during calibrator training to improve performance. The method achieves competitive calibration results on both in-distribution and out-of-distribution test sets compared to state-of-the-art methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper proposes a new way to make sure that machine learning models are confident only when they’re correct. Right now, many models just get more confident even if they’re wrong! The new method uses special versions of the sample data to train a “calibrator” that makes the model less confident when it’s wrong and more confident when it’s right. This approach is tested on different types of data and shows good results. |
Keywords
» Artificial intelligence » Cross entropy » Machine learning » Objective function
