Max Robotics — Global Service Robot Directory + US Certification

About

Nate Derbinsky is a cognitive systems researcher whose work focuses on memory management and learning within cognitive architectures, particularly the Soar architecture. His research addresses a critical challenge in artificial intelligence: how to maintain real-time performance and competence as an agent accumulates knowledge over long periods. Derbinsky’s most influential work, the 2011 paper “Performance evaluation of declarative memory systems in Soar” (16 citations), systematically evaluates how declarative memory scales under persistent learning, revealing computational bottlenecks that arise with knowledge growth. He further advanced the field by developing techniques for effective and efficient forgetting in working and procedural memories, as detailed in his 2013 paper (14 citations), and its companion 2012 work (4 citations) on competence-preserving retention. These contributions demonstrate that strategic knowledge removal can maintain agent performance without sacrificing learned capabilities. His more recent work (2021, 5 citations) extends these investigations into episodic memory, evaluating scaling challenges across multiple domains. Derbinsky’s research is foundational for building long-lived, autonomous agents that can learn continuously while remaining computationally tractable, making his work essential reading for anyone interested in cognitive architectures, memory systems, and lifelong learning in AI.