Here are some of the key open-source projects and tools I work on. You can also find all my code and repositories on my GitHub Profile.
ColoRL (TBA)
ColoRL is a deep reinforcement learning algorithm based on deep cross-entropy that generates lower bounds for graph theory conjectures, such as the Earth-Moon problem by finding counterexamples.
Bolt
Bolt is blazingly fast tool for learning Linear Temporal Logic (LTL) formulas from examples. This enables efficient glass-box interpretable machine learning on sequences, e.g. security logs.
Bolt is uses Boolean Set Cover as a core subroutine to combine LTL formulas using Boolean connectives, and is significantly faster (often over 100x speedup) than the state-of-the-art, while producing smaller models (formulas).
Repository: github.com/SynthesisLab/Bolt
GitHub Profile
Explore my other repositories, contributions, and research code on GitHub or the GitHub of my research team .