APPLICATIONS OF TECHNOLOGY:
- High-speed image recognition
- Energy-efficient neuromorphic computing, quantum computing
- Pattern recognition with error detection
- Low-power digital systems
BENEFITS:
- Significantly lower power consumption compared to traditional CMOS-based systems.
- Utilizes superconducting RSFQ technology for enhanced processing speed and energy efficiency.
- High-speed operation with frequencies up to tens of GHz.
BACKGROUND:
Neuromorphic computing aims to address limitations in traditional computing architectures, such as the gap between data processing and memory access speeds. Oscillatory Neural Networks (ONNs) are a promising approach, utilizing coupled oscillators to emulate neuron dynamics, offering potential for efficient image recognition. Current CMOS-based ONN implementations face challenges due to high power consumption and limited processing speeds. Superconducting technologies, like RSFQ, offer a solution with significantly lower power usage and faster operation, but require further development to overcome practical challenges such as cooling energy costs.
TECHNOLOGY OVERVIEW:
Scientists at Berkeley Lab have developed a superconducting oscillatory neural network (ONN) for image recognition, leveraging rapid single flux quantum (RSFQ) technology. This innovative system features inductively coupled ring oscillators based on Josephson junctions, which operate at frequencies up to tens of GHz (significantly faster than CMOS) while consuming only a few attojoules per operation, significantly less than CMOS systems. The ONN was also designed to perform pattern recognition by comparing a test pattern against a stored pattern using pairs of coupled oscillators.
DEVELOPMENT STAGE: Technology concept and/or application formulated.
PRINCIPAL INVESTIGATORS: Dilip Vasudevan
IP Status: Patent pending
Additional information: Superconducting-Oscillatory Neural Network With Pixel Error Detection for Image Recognition
OPPORTUNITIES: Available for licensing or collaborative research