BrainChip patents dynamic neural network model enabling edge biometrics and AI applicationsNov 21, 2019 |
Luana PascuCATEGORIES
Biometric R&D |
Biometrics News |
Biometrics StocksBrainChip has been granted United States Patent number 10,410,117 for the dynamic neural networks which are a valuable feature of its AI processing chip Akida, the company announced.
During a learning process, values are generated and stored in the synaptic registers of the AI device to generate a training model. Training models are themselves stored in the dynamic neural function library of the AI device, and the function library can then be used to train another device.
Through neural processing and memory access, Akida reduces computing resources required of the host CPU and cuts back on costs of running hyperscale data centers. Available as a licensable IP technology, it can be integrated into ASIC devices and will be available as an integrated SoC, and can be used for surveillance, advanced driver assistance systems (ADAS), autonomous vehicles (AV), vision guided robotics, drones, augmented and virtual reality (AR/VR), acoustic analysis, and Industrial Internet-of-Things (IoT).
Peter Van der Made, BrainChip founder and CTO with 40 years of experience in computer innovation, was recognized as inventor. Other inventions include a computer immune system at cybersecurity developer vCIS Technology, and a high resolution, high-speed color Graphics Accelerator Chip for IBM PC graphics. He authored the book ‘Higher Intelligence’ which talks about the architecture of the brain from a computer science point of view.
“This patent addresses efficiency that contributes to how Akida technology excels in speed, accuracy, and ultra-low power consumption,” said Van der Made. “Synapses store values, these thousands of synapses connect to thousands of neurons, and that neural output can be used by another set of neurons – which is closer to the way the human brain processes information.”
BrainChip holds a portfolio of 11 patents issued of pending, including one related to Spiking Neural Networks (SNN) that has been cited by leading companies such as IBM, Qualcomm, Samsung, and Hewlett Packard.
(20min delay)