Max Robotics — Global Service Robot Directory + US Certification

About

Sohrob Kazerounian is a researcher whose work sits at the intersection of artificial curiosity, neural dynamics, and autonomous learning systems. His key contributions center on developing intrinsic motivation frameworks and reinforcement learning algorithms that enable agents to autonomously build structured representations of their world. His most influential paper, "An intrinsic value system for developing multiple invariant representations with incremental slowness learning" (2013, 30 citations), introduces CD-MISFA, a model where artificial curiosity drives the formation of multiple stable sensory representations, simplifying complex inputs through incremental slowness learning. This work has been foundational for researchers exploring how agents can self-organize their learning without external rewards. Kazerounian has also made significant advances in neural dynamics, notably with his Dynamic Neural SARSA(λ) algorithm, which integrates Dynamic Field Theory with classical reinforcement learning to enable agents to learn behavioral sequences from delayed rewards. His research bridges computational neuroscience and robotics, offering frameworks that model human psychophysics while controlling robotic behavior. Though his citation counts are modest, his work represents a thoughtful integration of curiosity-driven learning and neural dynamics, appealing to researchers interested in autonomous development and cognitive architectures.