The difference between swarm learning and decentralized federated learning

[xiby]

Hello,
I am studying decentralized federated learning(DFL), and read your artical on nature. However, I found that DFL and swarm learning(SL) are so alike, such as they both train local model without leakage of private data, and they do not need a server to aggregate the global model. So I want to know what's the essential difference between SL and DFL.
Looking forward to your reply, thanks.

[sathya-SL]

Please refer papers or articles that talk about DFL for adding context to the discussion

[xiby]

I will provide some papers I have read:

1. Fully Decentralized Federated Learning
2. Peer-to-peer Federated Learning on Graphs
3. Blockchain Assisted Decentralized Federated Learning (BLADE-FL): Performance Analysis and Resource Allocation
4. BAFFLE : Blockchain Based Aggregator Free Federated Learning
5. D-Cliques: Compensating NonIIDness in Decentralized Federated Learning with Topology
6. A Blockchain-Based Decentralized Federated Learning Framework with Committee Consensus