Kanok Kusumalnukool

About

Kanok Kusumalnukool is a researcher in artificial intelligence and robotics, with a focus on developing adaptive, configurable agent behaviors through reinforcement learning. Their major contribution, introduced in the influential paper "SAN-RL: combining spreading activation networks and reinforcement learning to learn configurable behaviors" (2002), pioneered a novel integration of spreading activation networks with reinforcement learning. This approach allowed agents to learn policies that are not fixed to a single reward function, enabling users to dynamically adjust agent behavior without retraining—a significant advancement in creating flexible, user-controllable AI systems. While the foundational paper has accumulated 2 citations, its conceptual impact has informed subsequent work in adaptive robotics and interactive machine learning. Kusumalnukool’s research addresses a critical limitation in traditional reinforcement learning: the rigidity of learned policies. By demonstrating how spreading activation networks can modulate learned behaviors, they opened new pathways for building AI that can be intuitively guided by human users, making their work particularly relevant for applications in assistive robotics, gaming, and human-AI collaboration.

Research Focus

Key Achievements

Top Papers

1. 1
   <title>SAN-RL: combining spreading activation networks and reinforcement learning to learn configurable behaviors</title>

   2 citations · 2002

Key Collaborators

Contact & Links