News ·
KIPPO published at IJCAI 2025
Andrei Cozma and Landon Harris's Koopman-inspired take on proximal policy optimization brings dynamical-systems structure to reinforcement learning.
KIPPO: Koopman-Inspired Proximal Policy Optimization, by PhD students Andrei Cozma and Landon Harris with Dr. Qi, appeared at the International Joint Conference on Artificial Intelligence (IJCAI 2025).
KIPPO grew out of Andrei’s master’s thesis and marks the lab’s Koopman-theoretic research line reaching reinforcement learning: using Koopman-operator structure to regularize policy updates, improving stability and performance across control benchmarks.