METHOD FOR NEURAL NETWORK WITH WEIGHT QUANTIZATION: abstract simplified (17719294)

• This abstract for appeared for patent application number 17719294 Titled 'METHOD FOR NEURAL NETWORK WITH WEIGHT QUANTIZATION'

Simplified Explanation

This abstract describes a method for training and updating a spiking neural network (SNN) in two different devices. The method involves training the SNN in the first device to generate multiple weight values. These weight values are then used to calculate a set of second weight values based on a threshold value and the number of bits used for the weight values. The SNN is then retrained using these second weight values to update them. Finally, the updated weight values are saved in a memory in the second device for performing SNN operations.

Original Abstract Submitted

A method is provided and includes operations as below: training a spiking neural network (SNN) in a first device to generate multiple first weight values of M bits; calculating multiple second weight values of N bits corresponding to the first weight values according to a threshold value, the number M, and the first weight values, wherein the number N is smaller than the number M; retraining the spiking neural network with the second weight values to update the second weight values; and performing a write operation to save the updated plurality of second weight values in a memory in a second device for performing a spiking neural network operation in the second device.