---
id: "zkml"
name: "ZKML"
image: "zkml.webp"
section: "pse"
projectStatus: "inactive"
category: "research"
tldr: "ZKML (Zero-Knowledge Machine Learning) leverages zero-knowledge proofs for privacy-preserving machine learning, enabling model and data privacy with transparent verification."
tags:
  keywords: ["Anonymity/privacy", "Scaling"]
  themes: ["research"]
  types: ["Proof of concept", "Infrastructure/protocol"]
  builtWith: ["circom", "halo2", "nova"]
links:
  github: "https://github.com/socathie/circomlib-ml"
---

ZKML is a solution that combines the power of zero-knowledge proofs (ZKPs) and machine learning to address the privacy concerns in traditional machine learning. It provides a platform for machine learning developers to convert their TensorFlow Keras models into ZK-compatible versions, ensuring model privacy, data privacy, and transparent verification. ZKML can be used to verify if a specific machine learning model was used to generate a particular piece of content, without revealing the input or the model used. It has potential use cases in on-chain biometric authentication, private data marketplace, proprietary ML model sharing, and AIGC NFTs.