Techxplore January 21, 2022
The inherent optoelectronic noises make the photonic systems error-prone in practice. To mitigate and harness noises in photonic computing systems a team of researchers in the US (University of Washington University, Duke University, University of Maryland) has designed and demonstrated a photonic generative network as a part of a generative adversarial network (GAN). The network is implemented with a photonic core consisting of an array of programable phase-change memory cells to perform four-element vector-vector dot multiplication. The GAN can generate a handwritten number (“7”) in experiments and full 10 digits in simulation. They developed an optical random number generator, applied noise-aware training by injecting additional noise, and demonstrated the network’s resilience to hardware nonidealities. According to the researchers the results suggest the resilience and potential of more complex photonic generative networks based on large-scale, realistic photonic hardware…read more. Open Access TECHNICAL ARTICLE
Photonic GAN network with optoelectronic noises…. Credit: SCIENCE ADVANCES, 21 Jan 2022, Vol 8, Issue 3