Optical graph neural classification network and method based on on-chip diffraction neural network

Abstract

The invention relates to the technical field of optical neural network calculation, in particular to an optical graph neural classification network and method based on an on-chip diffraction neural network, and the network comprises an optical graph feature extraction module which is used for coding the input node attribute information of a graph structure to an input optical signal of an input waveguide, an output optical signal of the output waveguide propagated through the on-chip diffraction neural network is extracted, and graph feature information is obtained by using the output optical signal; the optical graph feature aggregation module is used for aggregating the graph feature information of the plurality of graph structures; and the classification module is used for classifying the plurality of graph structures according to the aggregated graph feature information of the plurality of graph structures to obtain a classification result of the plurality of graph structures. Therefore, the problem that an existing optical neural network can only process regular data structures in the forms of vectors, matrixes and the like and cannot process data structures of non-Euclidean spaces such as graph structures and the like is solved, and the on-chip integrated optical neural network can be utilized to process graph structure data such as social networks, paper intertraction networks and the like.

Description

technical field [0001] The present application relates to the technical field of optical neural network computing, in particular to an optical graph neural classification network and method based on an on-chip diffractive neural network. Background technique [0002] Deep learning techniques have made tremendous progress in a wide range of artificial intelligence (AI) applications, including computer vision, speech recognition, natural language processing, self-driving cars, biomedical science, and more. Its core is CPU (Central Processing Unit, central processing unit), GPU (Graphics Processing Unit, image processor), TPU (Tensor Processing Unit, tensor processor), FPGA (Field Programmable Gate Array, field programmable logic gate array. ), driven by the continuous development of comprehensive electronic computing platforms, using multi-layer neural networks to learn complex features from big data. However, with the increasing demand for artificial intelligence development...

AI Technical Summary

Problems solved by technology

However, how to effectively utilize optical computing to aid graph-based machine learning remains to be explored

Images