Denis Larionov

Papers

Total Citations

H-Index

About

Denis Larionov is a rising researcher at the forefront of efficient reinforcement learning, whose work bridges the gap between biological plausibility and real-world deployment. His primary research areas include spiking neural networks, neural network compression, and energy-efficient AI for robotics. Larionov’s major contributions are twofold: he pioneered a purely spiking approach to reinforcement learning, demonstrating that biologically-inspired, event-driven computation can achieve competitive performance in decision-making tasks—a breakthrough for low-latency, low-power systems. Complementing this, he has advanced neural network compression techniques specifically tailored for RL, showing how pruning and sparsity can dramatically reduce model size and energy consumption without sacrificing task accuracy. His most-cited works, each garnering 7 citations in their first year, are already shaping discussions on deploying RL in resource-constrained environments like autonomous robots. By tackling the critical challenges of inference speed and energy efficiency, Larionov is helping to make reinforcement learning practical for real-world applications where every millijoule and millisecond counts.