Summary of Selective Mixup Fine-tuning For Optimizing Non-decomposable Objectives, by Shrinivas Ramasubramanian et al.
Selective Mixup Fine-Tuning for Optimizing Non-Decomposable Objectives
by Shrinivas Ramasubramanian, Harsh Rangwani, Sho Takemori, Kunal Samanta, Yuhei Umeda, Venkatesh Babu Radhakrishnan
First submitted to arxiv on: 27 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); 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 paper addresses a pressing issue in machine learning, where massive amounts of data are generated and trained models must be evaluated on specific performance measures like worst-case recall and fairness constraints. Current state-of-the-art techniques fall short in achieving optimal performance on these practical objectives. To bridge the gap, the authors propose SelMix, a selective mixup-based fine-tuning technique for pre-trained models that optimizes for the desired objective. This framework determines a sampling distribution to perform feature mixing between samples from specific classes. The authors evaluate SelMix against existing empirical and theoretically principled methods on standard benchmark datasets for imbalanced classification, finding significant performance improvements across various non-decomposable objectives. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine having too much data! To use this data effectively, we need to make sure our machine learning models work well. But it’s not just about making the model work – we also want to make sure it’s fair and doesn’t favor one group over another. The problem is that current methods don’t always do a great job of achieving these goals. To fix this, researchers have come up with a new way called SelMix. It takes existing models and tweaks them so they perform better on specific tasks. This was tested on various datasets and showed significant improvements. |
Keywords
* Artificial intelligence * Classification * Fine tuning * Machine learning * Recall
