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

Diffractive optical neural networks have shown promising advantages overelectronic circuits for accelerating modern machine learning (ML) algorithms.However, it is challenging to achieve fully programmable all-opticalimplementation and rapid hardware deployment. Furthermore, understanding thethreat of adversarial ML in such system becomes crucial for real-worldapplications, which remains unexplored. Here, we demonstrate a large-scale,cost-effective, complex-valued, and reconfigurable diffractive all-opticalneural networks system in the visible range based on cascaded transmissivetwisted nematic liquid crystal spatial light modulators. With the assist ofcategorical reparameterization, we create a physics-aware training frameworkfor the fast and accurate deployment of computer-trained models onto opticalhardware. Furthermore, we theoretically analyze and experimentally demonstratephysics-aware adversarial attacks onto the system, which are generated from acomplex-valued gradient-based algorithm. The detailed adversarial robustnesscomparison with conventional multiple layer perceptrons and convolutionalneural networks features a distinct statistical adversarial property indiffractive optical neural networks. Our full stack of software and hardwareprovides new opportunities of employing diffractive optics in a variety of MLtasks and enabling the research on optical adversarial ML.

Physics-aware Complex-valued Adversarial Machine Learning in Reconfigurable Diffractive All-optical Neural Network