Summary of Logit Calibration and Feature Contrast For Robust Federated Learning on Non-iid Data, by Yu Qiao et al.
Logit Calibration and Feature Contrast for Robust Federated Learning on Non-IID Data
by Yu Qiao, Chaoning Zhang, Apurba Adhikary, Choong Seon Hong
First submitted to arxiv on: 10 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
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 In this paper, the authors explore the challenges of deploying federated learning models in edge networks, where data distribution is non-IID and vulnerable to adversarial examples. They highlight the limitations of directly applying adversarial training in federated learning, which can compromise accuracy. To address these issues, they propose FatCC, a method that combines local logit calibration and global feature contrast with vanilla federated adversarial training. This approach improves robust accuracy and clean accuracy in federated environments. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Federated learning is a way to train models on devices without sharing data. However, this can be tricky because devices have different kinds of data, and some bad actors might try to trick the system. The authors show that just using adversarial training (a technique to make models more robust) doesn’t work well in federated learning. Instead, they suggest FatCC, which helps make models both accurate and robust. |
Keywords
» Artificial intelligence » Federated learning
