Exploring on-chain smart contract use cases for zkML

Investigating zkML smart contract applications on blockchain.

By adding ML capabilities, smart contracts can become more autonomous and dynamic, allowing them to make decisions based on real-time on-chain data rather than static rules. ML capabilities will expand the automation, accuracy, efficiency, and flexibility of any smart contract on the chain. Cryptocurrency research firm 1kx explores the potential applications and use cases of on-chain ML and takes stock of emerging projects and progress of zkML core.

Verified reasoning scenarios have opened up a new design space for smart contracts. Some examples of native cryptographic applications include:

-DeFi: Verifiable off-chain machine learning oracles, machine learning parameterized DeFi applications, automated trading strategies;

– Security: Fraud monitoring for smart contracts;

– Traditional machine learning: Decentralized and trustless implementation of Kaggle, decentralized prompt markets for generating artificial intelligence;

– Identity: Using privacy-protecting biometric authentication instead of private keys, fair airdrops and contributor rewards;

– Web3 social: Filtering Web3 social media, advertising/recommendations;

– Creator economy/games: In-game economy rebalancing, new types of on-chain games.

Most of the focus of zkML is on the aforementioned infrastructure, but there are also some projects that focus on applications. Modulus Labs is one of the most diverse projects in the zkML field, dedicated to example applications and related research. Worldcoin is applying zkML to attempt to develop a privacy-preserving personality proof protocol. Giza can deploy AI models on-chain with fully trustless methods. Gensyn is a distributed hardware configuration network for training ML models. ZKaptcha focuses on bot problems in Web3, providing captcha services for smart contracts.

There are still some core challenges in the zkML space, including: minimizing accuracy loss in quantization; circuit size, especially when networks have multiple layers; effective proofs for matrix multiplication; and adversarial attacks.

Reference: https://mirror.xyz/1kx.eth/q0s9RCH43JCDq8Z2w2Zo6S5SYcFt9ZQaRITzR4G7a_k