Summary of Vertical Federated Learning with Missing Features During Training and Inference, by Pedro Valdeira and Shiqiang Wang and Yuejie Chi
Vertical Federated Learning with Missing Features During Training and Inference
by Pedro Valdeira, Shiqiang Wang, Yuejie Chi
First submitted to arxiv on: 29 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Distributed, Parallel, and Cluster Computing (cs.DC); Data Structures and Algorithms (cs.DS); Optimization and Control (math.OC)
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 This paper presents LASER-VFL, a novel approach for vertical federated learning that efficiently trains and inferences split neural network-based models while handling arbitrary sets of feature partitions. The proposed method relies on sharing model parameters and task-sampling to train a family of predictors. Unlike standard approaches, LASER-VFL can handle incomplete samples during training and supports inference even when some clients leave the federation after training. This is crucial for real-world scenarios where not all feature partitions are always available. The authors demonstrate that LASER-VFL achieves linear convergence rates and improved performance over baselines, including a 18.2% improvement in accuracy on CIFAR-100 when each of four feature blocks is observed with a probability of 0.5. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper talks about a new way to train artificial intelligence models that can work with incomplete data from different sources. Right now, most AI models need all the data at once, but this new method, called LASER-VFL, can handle missing pieces of information and still make good predictions. This is important because in real life, not everyone has access to the same information or data all the time. The researchers show that their new method works better than older approaches and can even improve accuracy by 18% on a specific task. |
Keywords
» Artificial intelligence » Federated learning » Inference » Neural network » Probability
