Photon neural network chip

Abstract

The embodiment of the invention provides a neural network chip. The neural network chip comprises an optical input module, an optical processing module and an optical output module, wherein the optical input module at least comprises a wavelength division multiplexer which is used for receiving a plurality of paths of input optical signals and multiplexing the plurality of paths of input optical signals into one path of first target optical signal. And the optical processing module is connected with the output end of the optical input module and is used for calculating the first target optical signal to generate a second target optical signal. And the optical output module is connected with the output end of the optical processing module, at least comprises a wavelength division demultiplexer and is used for decomposing the second target optical signal into multiple paths of output optical signals. Visibly, according to the scheme, the optical signals are subjected to wavelength division multiplexing, and the multiple optical signals are input into the optical processing module through the light beams with different wavelengths, so that the single-path optical propagation information capacity is increased, and the calculation speed of the photonic neural network is increased.

Description

technical field [0001] The invention relates to wavelength division multiplexing technology, in particular to a photonic neural network chip. Background technique [0002] At present, the artificial neural algorithm operated by the neural network chip includes a large number of matrix multiplication operations, and the computing platform of the traditional computer architecture (such as the computing platform composed of CPU or GPU) needs a large computing power when processing these operations. support. Specifically, it is necessary to use the calculation module to learn different operation logics in depth. However, the versatility of many operation logics cannot be completed efficiently, which in turn leads to low calculation efficiency. [0003] Therefore, how to provide a neural network chip capable of improving data calculation efficiency is a major technical problem to be solved urgently by those skilled in the art. Contents of the invention [0004] In view of thi...

AI Technical Summary

Problems solved by technology

Specifically, it is necessary to use the calculation module to learn different operation logics in depth. However, the versatility makes many operation logics cannot be completed efficiently, which leads to low calculation efficiency.

Images