Golden Grain: Building a Secure and Decentralized Model Marketplace for MLaaS

ML-as-a-service (MLaaS) becomes increasingly popular and revolutionizes the lives of people. A natural requirement for MLaaS is, however, to provide highly accurate prediction services. To achieve this, current MLaaS systems integrate and combine multiple well-trained models in their services. However, in reality, there is no easy way for MLaaS providers, especially for startups, to collect well-trained models from individual developers, due to the lack of incentives. In this paper, we aim to fill this gap by building a model marketplace, called as GoldenGrain, to facilitate model sharing, which enforces the fair model-money swaps between individual developers and MLaaS providers. Specifically, we deploy the swapping process on the blockchain, and further introduce a blockchain-empowered model benchmarking design for transparently determining the model prices according to their authentic performances so as to incentivize the faithful contributions of well-trained models. Especially, to ease the blockchain overhead for benchmarking, our marketplace carefully offloads the heavy computation and crafts a trusted execution environment (TEE) based secure off-chain on-chain interaction protocol, ensuring both the integrity and authenticity of benchmarking. We implement a prototype of our GoldenGrain on the Ethereum blockchain, and extensive experiments with standard benchmark datasets demonstrate the practically affordable performance of our design.