One of the problems with large machine learning models is well there's two problems. So for example the weights of GPT-3 there's 175 billion of them and computing to that is very very expensive. There's actually a lot of computation that happens internally within the model and representing that within ZK is also very expensive. The basic problem is that the proof systems that are really good for matrix multiplication aren't good at nonlinearities and vice versa. We're exploring how to mitigate in terms of the performance implications of that.
This week, Anna Rose and Tarun Chitra dive back into the topic of ZK ML with guests Yi Sun, co-founder of Axiom, and Daniel Kang, Assistant Professor of computer science at UIUC. They discuss Yi and Daniel’s previous academic work and what led them to get interested in ZK topics and specifically ZK ML. They then dive into a discussion about 2 recent papers which examine the use of ZK within Machine Learning architectures.
Here are some additional links for this episode:
Apply for ZK Hack Lisbon here: ZK Hack application
Aleo is a new Layer-1 blockchain that achieves the programmability of Ethereum, the privacy of Zcash, and the scalability of a rollup.
Interested in building private applications? Check out Aleo’s programming language called Leo by visiting http://developer.aleo.org.
You can also participate in Aleo’s incentivized testnet3 by downloading and running a snarkOS node. No sign-up is necessary to participate.
For questions, join their Discord at aleo.org/discord.
If you like what we do: