Summary of Communication-efficient Federated Knowledge Graph Embedding with Entity-wise Top-k Sparsification, by Xiaoxiong Zhang et al.
Communication-Efficient Federated Knowledge Graph Embedding with Entity-Wise Top-K Sparsification
by Xiaoxiong Zhang, Zhiwei Zeng, Xin Zhou, Dusit Niyato, Zhiqi Shen
First submitted to arxiv on: 19 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
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 proposes a new approach to federated knowledge graph embedding learning, addressing the issue of large parameter sizes and extensive communication rounds in existing methods. The authors find that reducing the size of parameters transmitted within each communication round can significantly improve convergence speed. They propose a bidirectional communication-efficient method, FedS, which uses entity-wise top-K sparsification to identify and upload only the most important entity embeddings during compression. The server then performs personalized embedding aggregation for each client and transmits the top-K aggregated embeddings. Additionally, an intermittent synchronization mechanism is used to mitigate the negative effects of embedding inconsistency among shared entities. The authors demonstrate the effectiveness of FedS across three datasets, showing significant improvements in communication efficiency with minimal performance degradation. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper tries to make federated learning more efficient by reducing the amount of information that needs to be sent between devices. Right now, each device has to send a lot of data and this can slow things down. The authors found that if they reduce the size of the data being sent, it can actually help things run faster. They came up with a new way to do this called FedS, which is like a filter that only sends the most important information. They also added a special feature to make sure all the devices are in sync, so everything works smoothly. The authors tested their method on three different sets of data and it worked really well, making things faster without sacrificing accuracy. |
Keywords
* Artificial intelligence * Embedding * Federated learning * Knowledge graph
