Flock.io Partners with Sindri Labs to Leverage Zero-Knowledge Proofs for Model Update Validity

Files & media

In today's digitally scrutinized landscape, FLock.io, a decentralized model training network and leading innovator in the field of Federated Learning (FL), is joining forces with Sindri Labs, a startup transforming complex zero-knowledge proving infrastructure into developer-friendly APIs to provide serverless, on-demand proof generation at scale.  The partnership introduces an innovative twist, integrating zero-knowledge proofs (ZKPs) to safeguard against vulnerabilities like malicious exploitation by central aggregators. This ensures swift and reliable model updates without compromising data integrity or user trust. It's not just about setting a higher standard – it's about redefining the norms.

At the heart of this collaboration is a clear-cut mission: enhancing the trustworthiness of the FL system by countering potential malicious actions from aggregators, such as abandoning and replacing clients' training models or launching Sybil attacks with fake clients. With ZKPs in our corner, FLock.io transforms the FL aggregator into a reliable proof-producing machine, validating faithful execution without exposing confidential training information. ZKPs stand out for their unique capability to verify proof accuracy while maintaining data privacy for external entities that are not part of the FL system. This fusion of technology and trust adds a game-changing layer of security, safeguarding against threats.

"Federated learning has incredible potential to drive innovation across industries,  but its broader adoption requires robust measures to ensure the integrity and trust of the process," commented Jiahao Sun,  the CEO of FLock.io. "Our partnership with Sindri Labs marks a significant stride towards achieving this goal, utilizing zero-knowledge proofs to establish a secure and transparent environment for all stakeholders."

“We got tired of seeing fantastic ZK developer teams weighed down by devops and infrastructure required to generate proofs at scale; so we decided to do something about it,” said Stasia Carson, co-founder of Sindri Labs. “Sindri allows developer teams like FLock.io to deploy scalable ZK apps in minutes, without the headache of managing a complicated infrastructure stack. We’re thrilled to have an opportunity to put this to the test and partner with a team like FLock.io as they continue to push the boundaries of federated learning.”

FLock.io develops decentralized and privacy-preserving federated learning , builds an on-chain ML fine-tuning and inferencing network, so that the community can create models without handing over data ownership to big tech. Sindri Labs is transforming the bottlenecks, overhead and roadblocks facing ZK developer teams by simplifying proof generation for ZK applications, L2s and zkVMs.

FLock.io with the help of Sindri Labs, brings top-tier know-how to provide robust decentralized federated learning resulting in an unprecedented game-changing solution.  It not only combats malicious actions during model aggregation but also furnishes clients with a verifiable and transparent system.