[3PS-039]
On-Chip Hebbian Learning with an Integrated Neuromorphic Device Platform
발표자이정화 (연세대학교)
연구책임자김선권 (연세대학교 화공생명공학과), 조정호 (연세대학교 화공생명공학과)
Abstract
Conventional architectures face von Neumann bottleneck and surging AI energy demands. While neuromorphic computing addresses these limitations, efficient on-chip learning remains challenging. Here we present an integrated neural platform combining modulation-optimized presynaptic transistors, threshold switching memristor neurons, and adaptive feedback synapses. This system implements real-time Hebbian learning through correlation-based synaptic weight updates without complex peripheral circuitry. Comprehensive characterization of a 6×6 array confirms stable device operation and successful execution of local learning rules. Experimental results demonstrate clear correlation between input-output patterns and subsequent weight modifications, validating hardware implementation of Hebbian plasticity and establishing a pathway for autonomous on-chip learning in neuromorphic systems.