Democratizing Federated Learning with Blockchain and Multi-Task Peer Prediction
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910085325783040 |
|---|---|
| author | Witt, Leon Toyoda, Kentaroh Samek, Wojciech Li, Dan |
| author_facet | Witt, Leon Toyoda, Kentaroh Samek, Wojciech Li, Dan |
| contents | The synergy between Federated Learning and blockchain has been considered promising; however, the computationally intensive nature of contribution measurement conflicts with the strict computation and storage limits of blockchain systems. We propose a novel concept to decentralize the AI training process using blockchain technology and Multi-task Peer Prediction. By leveraging smart contracts and cryptocurrencies to incentivize contributions to the training process, we aim to harness the mutual benefits of AI and blockchain. We discuss the advantages and limitations of our design. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_28434 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Democratizing Federated Learning with Blockchain and Multi-Task Peer Prediction Witt, Leon Toyoda, Kentaroh Samek, Wojciech Li, Dan Cryptography and Security Computers and Society The synergy between Federated Learning and blockchain has been considered promising; however, the computationally intensive nature of contribution measurement conflicts with the strict computation and storage limits of blockchain systems. We propose a novel concept to decentralize the AI training process using blockchain technology and Multi-task Peer Prediction. By leveraging smart contracts and cryptocurrencies to incentivize contributions to the training process, we aim to harness the mutual benefits of AI and blockchain. We discuss the advantages and limitations of our design. |
| title | Democratizing Federated Learning with Blockchain and Multi-Task Peer Prediction |
| topic | Cryptography and Security Computers and Society |
| url | https://arxiv.org/abs/2603.28434 |