Contributed examples¶
Our Optical Processing Unit has already been used in publications. The open source code can be found in the following repositories:
NEWMA: a new method for scalable model-free online change-point detection¶
Code: https://github.com/lightonai/newma
Paper: https://ieeexplore.ieee.org/abstract/document/9078835
Au Revoir Backprop! Bonjour Optical Transfer Learning!¶
Code: https://github.com/lightonai/transfer-learning-opu
Blog post: https://medium.com/@LightOnIO/au-revoir-backprop-bonjour-optical-transfer-learning-5f5ae18e4719
Beyond Overfitting and Beyond Silicon: The double descent curve¶
Accelerating SARS-COv2 Molecular Dynamics Studies with Optical Random Features¶
Code: https://github.com/lightonai/newma-md
Blog post #1: https://medium.com/@LightOnIO/accelerating-sars-cov2-molecular-dynamics-studies-with-optical-random-features-b8cffdb99b01
Blog post #2: https://medium.com/@LightOnIO/optical-random-features-versus-sars-cov-2-glycoprotein-trajectory-round-2-adcf04d6036d
Technical report: https://arxiv.org/abs/2006.08697
Kernel computations from large-scale random features obtained by Optical Processing Units¶
Don’t take it lightly: Phasing optical random projections with unknown operators¶
Code: https://github.com/swing-research/opu_phase
Paper: https://proceedings.neurips.cc/paper/2019/hash/b49fdab097253cac48e3dc628a49da5e-Abstract.html
Fast graph classifier with optical random features¶
Code: https://github.com/hashemghanem/OPU_Graph_Classifier
Paper: https://arxiv.org/abs/2010.08270 (to appear at ICASSP 2021)
Meetup video: https://youtu.be/vKAVamGH6wg
Adversarial Robustness by Design through Analog Computing and Synthetic Gradients¶
Code: https://github.com/lightonai/adversarial-robustness-by-design