Build a project that combines zk technology and machine learning. You might consider designing a protocol or a piece of a protocol for one of the following: - Proving correct inference of a machine learning model. - Proving correct training of a model. - Federated learning. - Others: surprise us with your creativity! The list of references below is not exhaustive by any means, we also encourage you to look for your own references.

Considering security from the beginning of your project is the best way to ensure your code is secure. Your project should demonstrate that security was a forethought during development, featuring high-quality test suites and use of security tooling.

This prize spotlights innovations that aim to address, optimise or innovate in the realm of Preconfirmations on Ethereum. We encourage diverse projects, from theoretical explorations to practical applications, with a particular emphasis on Preconfirmations. We invite participants to draw inspiration from the pioneering work of Flashbots, SUAVE, and ongoing Ethereum community research. Suggested areas of focus include: - Inclusion Preconfirmations: Solutions that enhance Ethereum's transaction ordering process, with a focus on inclusion preconfirmation mechanisms. - Pricing Preconfirmations: Develop solutions for efficient and transparent pricing in preconfirmation processes. - Data Dashboards: Development of tools and dashboards that provide insights into preconfirmation activities across existing infrastructure. - Research Applications: Theoretical or applied research projects that contribute to the understanding and effective use of preconfirmations.

Starknet provides developers with really low fees. Another way to look at low fees is “more compute”. Create a Starknet-based protocol that uses significantly more gas than Ethereum applications to achieve something previously impossible on Ethereum. There are two obvious ways to think about this problem: 1. Now that fees are cheap, will protocol participants be willing to pay fractions of a cent more to do the ‘housekeeping’ for a protocol: 1. Rebalance other users’ positions in your protocol 2. Reduce back-running arbitrage by solving for it onchain 3. Trigger abstract cron jobs with left over gas 2. Some problems have high-complexity classes that may now be executable on the chain at a reasonable price. 1. Some intractable problems have reasonable heuristic or stochastic solutions that may now be possible. 2. Large-scale Monte Carlo simulation may be possible (randomness is available from the Pragma oracle) 3. Problems in NP may be achievable for a small number of elements. For example: 1. Closing back-running arbitrage may be quite expensive to compute, but it may still be worth doing a naive strategy on-chain up to a certain level of gas consumption. 2. Traveling salesman for a tiny number of accounts may be possible.