We use cookies to offer a better browsing experience, analyze site traffic, personalize content, and serve targeted advertisements. By clicking accept, you consent to our privacy policy & use of cookies. (Privacy Policy)

Nil Foundation Teams with Taceo to Revolutionise Machine Learning

Nil Foundation, a zero-knowledge proofs marketplace, has teamed up with the zero-knowledge solutions provider Taceo to build a platform for validating Machine Learning (ML) models on Layer-1 blockchains. By implementing zero-knowledge verification on the Ethereum Layer 1 mainnet, the two firms want to make trustless machine learning a reality.

What this means for artificial intelligence

The partnership employs the zkLLVM compiler tool developed by the Nil Foundation. The compiler tool can verify ML computations in a wide range of popular programming languages, paving the way for a future in which ML can be used safely within smart contracts on the blockchain.

Machine learning is a subfield of AI that enables computers to improve their performance on tasks through the acquisition of knowledge and skills through repeated exposure to experience and responses. The necessity for developers to rebuild models in zero-knowledge domain-specific languages (zkDSL) is removed thanks to zkLLVM.

Taceo points out that, for GPT-like models in particular, it can be difficult to prove the accurate execution of existing Neural Networks (NNs) to entities like Ethereum. Moreover, developers are few because creating NN circuits often necessitates familiarity with ZK technology. The company intends to use the LLVM compiler toolchain to automatically construct circuits from trained NNs.

The major goal of this collaboration is to create a secure infrastructure for the validation of ML models on the Ethereum blockchain. This ground-breaking work intends to make it possible for ML apps to work within smart contracts without the requirement for trusted third parties.

According to Taceo, zero-knowledge proofs (zkps) within Ethereum smart contracts are necessary to prove the execution of ML operations and the integrity of training datasets. The necessity for machine learning powered decision making over enormous datasets is becoming more obvious as the number of decentralised applications (dApps) grows. But Teceo empasises that ML models need to be proven so that these programmes can run safely.