Cristian Bodnar

Papers

Total Citations

H-Index

About

Cristian Bodnar is a rising star in machine learning whose work bridges reinforcement learning, geometric deep learning, and graph neural networks. His most cited paper, "Proximal Distilled Evolutionary Reinforcement Learning" (2020, 71 citations), introduced a novel hybrid framework that synergizes evolutionary algorithms with deep reinforcement learning, enabling more sample-efficient and robust policy optimization in complex environments. This work demonstrated how genetic algorithms could be effectively scaled to deep neural networks, challenging conventional wisdom about their limitations. Beyond this, Bodnar has made significant contributions to geometric deep learning, particularly in developing message-passing frameworks for graphs and manifolds that respect underlying symmetries. His research on equivariant and invariant architectures has advanced the theoretical foundations of learning on non-Euclidean domains. With growing citation impact and recognition in top venues like NeurIPS and ICLR, Bodnar is establishing himself as a leading voice in the next generation of AI researchers, pushing the boundaries of how machines can learn from structured data and complex environments.