Summary of Don’t Forget What I Did?: Assessing Client Contributions in Federated Learning, by Bishwamittra Ghosh et al.
Don’t Forget What I did?: Assessing Client Contributions in Federated Learning
by Bishwamittra Ghosh, Debabrota Basu, Fu Huazhu, Wang Yuan, Renuga Kanagavelu, Jiang Jin Peng, Liu Yong, Goh Siow Mong Rick, Wei Qingsong
First submitted to arxiv on: 11 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
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 Federated Learning (FL) framework, called FLContrib, addresses the challenge of fairly assessing client contributions in collaborative machine learning without exposing private data. Existing methods rely on simplified assumptions and cooperative game-theoretic concepts like Shapley values. The novel history-aware approach integrates the FL training process and linearity of Shapley value to provide a timeline of client contributions over epochs. To optimize computations, a scheduling procedure is introduced, balancing correctness and efficiency while considering two-sided fairness criteria. Experimental results demonstrate a controlled trade-off between assessment accuracy and computational cost. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Federated Learning lets many devices work together to create an AI model without sharing their private data. One challenge is making sure each device gets the right credit for contributing to the model. Most methods use special math from game theory, but they make simplifying assumptions. The new FLContrib method combines these math concepts with information about how the training process works and uses it to give a timeline of how much each device contributed over time. To save computer power, FLContrib also has a way to decide which steps in the training process are most important to compute accurately. This helps balance accuracy and speed while being fair to all devices. |
Keywords
* Artificial intelligence * Federated learning * Machine learning
