A neuromorphic processor with on-chip learning for beyond-CMOS device integration

Area of Science:

• Neuromorphic Engineering
• Materials Science
• Computer Architecture

Background:

• Emerging memory technologies offer potential for neuromorphic systems but face integration challenges.
• A gap exists between material development and large-scale, functional neuromorphic system realization.
• Selecting optimal devices and materials for specific functions and CMOS integration is crucial.

Purpose of the Study:

• To present a mixed-signal neuromorphic architecture for exploring on-chip learning and novel device integration.
• To serve as a platform bridging silicon-based neuromorphic computation and emerging devices.
• To demonstrate the architecture's readiness for device integration through testing and simulation.

Main Methods:

• Development of a mixed-signal neuromorphic architecture.
• Integration of on-chip learning circuits and novel two- and three-terminal devices.
• Comprehensive measurements and simulations to validate device integration readiness.

Main Results:

• Demonstrated the architecture's capability to integrate emerging devices.
• Validated the platform's suitability for testing bio-inspired learning algorithms.
• Established a tangible link between brain-inspired computation and device research.

Conclusions:

• The presented architecture is ready for integrating novel memory devices into neuromorphic systems.
• The platform facilitates the practical testing of emerging devices with bio-inspired algorithms.
• This work bridges the gap between advanced materials and functional neuromorphic electronic systems.