Two frontier fields meet at a crossroad: Zero-Knowledge Proofs (ZKPs) and Machine Learning (ML). The confluence of cryptographic advancements and increasingly data-driven decisions may revolutionise how we handle privacy, security, and intelligence in the digital age. ML thrives on vast datasets, fueling advancements in AI that dictate everything from financial forecasts to medical diagnostics. Simultaneously, ZKPs, a cryptographic technique, allow one party to prove to another that a statement is true, without conveying any specific information, thus preserving data privacy and safety.

ML models can be trained on sensitive data without ever accessing the data itself. This marks a leap towards robust data privacy without compromising the computational integrity and intelligence that ML offers. ML models could be trained with more data and more sensitivity at the same time. We will begin with a primer on ZKPs and ML, setting the stage for a deep dive into integration challenges, research areas, and real-world applications, from healthcare & finance to deepfakes & surveillance.

This lecture will be led by Manuj Mishra, who built ZK Microphone at ETH CC and is actively building in ZKML, with a focus on provenance.