English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Photonic neural networks perform brain-inspired computations using photonsinstead of electrons that can achieve substantially improved computingperformance. However, existing architectures can only handle data with regularstructures, e.g., images or videos, but fail to generalize to graph-structureddata beyond Euclidean space, e.g., social networks or document co-citationnetworks. Here, we propose an all-optical graph representation learningarchitecture, termed diffractive graph neural network (DGNN), based on theintegrated diffractive photonic computing units (DPUs) to address thislimitation. Specifically, DGNN optically encodes node attributes into stripoptical waveguides, which are transformed by DPUs and aggregated by on-chipoptical couplers to extract their feature representations. Each DPU comprisessuccessive passive layers of metalines to modulate the electromagnetic opticalfield via diffraction, where the metaline structures are learnable parametersshared across graph nodes. DGNN captures complex dependencies among the nodeneighborhoods and eliminates the nonlinear transition functions during thelight-speed optical message passing over graph structures. We demonstrate theuse of DGNN extracted features for node and graph-level classification taskswith benchmark databases and achieve superior performance. Our work opens up anew direction for designing application-specific integrated photonic circuitsfor high-efficiency processing of large-scale graph data structures using deeplearning.

All-optical graph representation learning using integrated diffractive photonic computing units