TY - GEN
T1 - Smart Contract Aided Dynamic Multi-Owner Copyright with Hierarchical Differential Royalties for Collaborative AI Model Trading
AU - Fan, Yixin
AU - Wu, Jun
N1 - Publisher Copyright:
© 2024 IEEE.
PY - 2024
Y1 - 2024
N2 - Currently, Artificial Intelligence (AI) model trading has gained increasing attention from academia and industry. Although of great appeal, it poses a series of challenges when applied in decentralized collaborative trading scenarios. On the one hand, model owners should constantly fine-tune AI models to meet customized needs, which causes heavy updating overhead. On the other hand, collaborative AI model training results in copyright sharing between multiple owners, which necessitates differential AI model royalties to adapt to the development of trading market. To address these challenges, we design a smart contract aided dynamic multi-owner copyright scheme with hierarchical differential royalties for collaborative AI model trading. First, to meet the frequent updating requirements, we propose a Non-Fungible Token (NFT) driven dynamic AI model copyright protection structure, integrating InterPlanetary File System (IPFS) and NFT to realize low-cost AI model updating. Second, a batch AI model trading smart contract is designed to further reduce the trading cost in high-frequency trading scenarios. Third, we establish a hierarchical differential royalties collection and allocation mechanism for multi-owner AI models, where the AI model royalties are adaptively adjusted over time, utilization frequency, and the number of customers, and are allocated differentially based on training contributions. Last but not least, we conduct comprehensive evaluation and analysis of our scheme, demonstrating its efficiency, availability and flexibility.
AB - Currently, Artificial Intelligence (AI) model trading has gained increasing attention from academia and industry. Although of great appeal, it poses a series of challenges when applied in decentralized collaborative trading scenarios. On the one hand, model owners should constantly fine-tune AI models to meet customized needs, which causes heavy updating overhead. On the other hand, collaborative AI model training results in copyright sharing between multiple owners, which necessitates differential AI model royalties to adapt to the development of trading market. To address these challenges, we design a smart contract aided dynamic multi-owner copyright scheme with hierarchical differential royalties for collaborative AI model trading. First, to meet the frequent updating requirements, we propose a Non-Fungible Token (NFT) driven dynamic AI model copyright protection structure, integrating InterPlanetary File System (IPFS) and NFT to realize low-cost AI model updating. Second, a batch AI model trading smart contract is designed to further reduce the trading cost in high-frequency trading scenarios. Third, we establish a hierarchical differential royalties collection and allocation mechanism for multi-owner AI models, where the AI model royalties are adaptively adjusted over time, utilization frequency, and the number of customers, and are allocated differentially based on training contributions. Last but not least, we conduct comprehensive evaluation and analysis of our scheme, demonstrating its efficiency, availability and flexibility.
KW - AI model trading
KW - blockchain
KW - Copyright protection
KW - smart contract
UR - http://www.scopus.com/inward/record.url?scp=85198035772&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85198035772&partnerID=8YFLogxK
U2 - 10.1109/SmartCloud62736.2024.00021
DO - 10.1109/SmartCloud62736.2024.00021
M3 - Conference contribution
AN - SCOPUS:85198035772
T3 - Proceedings - 2024 IEEE 9th International Conference on Smart Cloud, SmartCloud 2024
SP - 77
EP - 81
BT - Proceedings - 2024 IEEE 9th International Conference on Smart Cloud, SmartCloud 2024
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 9th IEEE International Conference on Smart Cloud, SmartCloud 2024
Y2 - 10 May 2024 through 12 May 2024
ER -