Break the Von Neumann Architecture completely
Third generation of SNN, spiking neural network
The spiking neural network with complete algorithm is based on STDP(spike-time-dependent plasticity)
Brain-inspired chips implement electric circuit to simulate spiking neural network. The chips can simulate as much as 50 million neurons by the dynamic distribution way.
The neuromorphic chip can teach, train and enhance itself with high level of integration, of which template is easy to change, and eventually solve complex artificial intelligence problem.
The power consumption is only a small percentage of the consumption from traditional chips.
Break the Von Neumann Architecture completely
A general intelligence chip to process pattern recognition
Can learn and adjust the priority between neurons based on the OPIUM lite.
here are 12.8 million neurons in one chip and can be distributed by specialized instructions.
It is the first time to achieve Binary Neural Network by on-chip online learning.
It is much more efficient to learn and recognize than running on general hardware such as CPU, GPU in a traditional way such as CNN while the consumption is much lower.