Summary of Robust Model Evaluation Over Large-scale Federated Networks, by Amir Najafi et al.
Robust Model Evaluation over Large-scale Federated Networks
by Amir Najafi, Samin Mahdizadeh Sani, Farzan Farnia
First submitted to arxiv on: 26 Oct 2024
Categories
- Main: Machine Learning (stat.ML)
- Secondary: Machine Learning (cs.LG)
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 A machine learning model’s performance on an unseen target network is certified using measurements from an available source network, where heterogeneous datasets are distributed across clients connected to a central server. The goal is to provide guarantees for the model’s performance on another, different target network, assuming the deviation between meta-distributions is bounded by Wasserstein distance or f-divergence. Theoretical guarantees are derived for empirical average loss and uniform bounds are established on risk CDF, which correspond to novel and adversarially robust versions of Glivenko-Cantelli theorem and DKW inequality. These bounds can be computed in polynomial time with a polynomial number of queries to clients, preserving client privacy. Non-asymptotic generalization bounds consistently converge to zero as both client sample size and total clients increase. The paper’s results are validated through extensive empirical evaluations across real-world tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A team of researchers has found a way to ensure that machine learning models work well on new data, even if the training data is different from what they’ve seen before. They used measurements from a network where lots of devices share data with each other. The goal was to make sure the model would perform well on a brand new network, as long as there wasn’t too much difference between the two networks. The team developed mathematical formulas to prove that their method works and tested it on real-world problems. |
Keywords
» Artificial intelligence » Generalization » Machine learning
