Nil Foundation collaborates with Taceo for machine learning model verification on Ethereum

Quick Take

The Nil Foundation and Taceo are collaborating on a pipeline for validating ML models on Ethereum’s Layer 1 mainnet using zero-knowledge techniques.
The main goal is to make ML models verifiable on the Ethereum blockchain, eliminating the need for third-party trust in smart contracts.

The Nil Foundation and research firm Taceo are collaborating to create a software pipeline that focuses on the validation of machine learning models on Ethereum’s Layer 1 mainnet using zero-knowledge techniques.

The primary objective of this collaboration is to make ML models provable on the Ethereum blockchain. This would enable ML operations within smart contracts without the need for third-party trust.

Machine learning is a branch of artificial intelligence that allows computers to enhance task performance by gaining experience, usually by examining data and detecting patterns.

A pivotal element of this joint venture is the Nil Foundation’s zkLLVM. This compiler tool is tailored to validate data within software systems without reliance on trust, utilizing ZK proofs. The compiler can authenticate machine learning computations in various mainstream development languages supported by LLVM, including C++, Rust, and JavaScript/TypeScript. Such functionality is useful in the realm of blockchain and cryptographic applications, where there’s a necessity to verify computations without disclosing data.

The Nil Foundation first launched zkLLVM in February, allowing developers to employ zero-knowledge techniques in familiar coding languages.

Verifying computations without revealing data

Including zero-knowledge proofs offers the potential to verify ML operations and their associated training datasets. Such a feature allows for the integration of verifiable ML into decentralized applications, impacting various sectors — including DeFi, privacy, and emerging areas such as healthcare.

Taceo and Nil Foundation are working on developing the pipeline in a testing phase and will release the initial outcomes and usability of zkLLVM within ML models in Q4 2023.

“The integration of machine learning into decentralized applications requires that ML models are both secure and provable, especially when interfacing with smart contracts on platforms like Ethereum,” Taceo said in a statement.

In January 2023, the Nil Foundation raised a $22 million round led by Polychain Capital.