Summary of Codream: Exchanging Dreams Instead Of Models For Federated Aggregation with Heterogeneous Models, by Abhishek Singh et al.
CoDream: Exchanging dreams instead of models for federated aggregation with heterogeneous models
by Abhishek Singh, Gauri Gupta, Ritvik Kapila, Yichuan Shi, Alex Dang, Sheshank Shankar, Mohammed Ehab, Ramesh Raskar
First submitted to arxiv on: 25 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 CoDream is a novel framework that extends Federated Learning (FL) by aggregating “knowledge” derived from models, rather than model parameters. This approach enables collaborative optimization of randomly initialized data using federated optimization in the input data space. The key insight is that jointly optimizing this data can effectively capture the properties of the global data distribution. CoDream offers numerous benefits, including model-agnostic collaborative learning, independent communication, compatibility with secure aggregation, and adaptive optimization for personalized learning. Our code is available at https://mitmedialab.github.io/codream.github.io/. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary CoDream is a new way to learn from lots of different computers without sharing their models. Normally, when we want computers to work together on a task, we share the model they use and have them adjust it based on their own data. But with CoDream, we’re doing something different. We’re sharing knowledge about what’s important in the data itself, rather than the models. This helps computers learn from each other without needing to agree on what kind of model to use. It also makes it easier for computers with different models to work together and keeps their data private. |
Keywords
* Artificial intelligence * Federated learning * Optimization
