Summary of Cdfl: Efficient Federated Human Activity Recognition Using Contrastive Learning and Deep Clustering, by Ensieh Khazaei et al.
CDFL: Efficient Federated Human Activity Recognition using Contrastive Learning and Deep Clustering
by Ensieh Khazaei, Alireza Esmaeilzehi, Bilal Taha, Dimitrios Hatzinakos
First submitted to arxiv on: 17 Jul 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: 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 The proposed paper presents an efficient federated learning framework for image-based Human Activity Recognition (HAR) called CDFL. The framework addresses challenges in traditional machine learning approaches, such as memory-intensive processing on central servers and privacy concerns. By training a global model collaboratively across multiple devices using local model parameters, FL can improve performance and reduce data sharing. However, realistic settings often involve non-independently and identically distributed (Non-IID) sensor data, leading to slow convergence and poor performance. CDFL addresses these challenges by selecting representative privacy-preserved images using contrastive learning and deep clustering, reducing communication overhead, and improving global model quality. Experiments on three public datasets demonstrate the superiority of CDFL in terms of performance, convergence rate, and bandwidth usage. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about a new way to recognize human activities using sensors from different devices. Right now, machine learning approaches need lots of data and can be slow and wasteful. Federated Learning (FL) helps solve this problem by having devices share their local models instead of data. However, in real-life situations, the sensor data on each device is not always the same, which makes it hard for FL to work well. This paper proposes a new framework called CDFL that can handle these issues and improve performance. It does this by selecting important images, reducing the amount of information shared between devices, and making sure the global model is accurate. The results show that CDFL works better than other approaches in terms of how well it recognizes activities. |
Keywords
» Artificial intelligence » Activity recognition » Clustering » Federated learning » Machine learning
