Experimental Quantum Stochastic Walks Simulating Associative Memory of Hopfield Neural Networks
Hao Long Tang, Zhen Feng, Yinghan Wang, Peng-Cheng Lai, Chaoyue Wang, Zhuoyang Ye, Cheng-Kai Wang, Zi-Yu Shi, Tianyu Wang, Yuan Chen, Jun Gao, Xian‐Min Jin · Physical Review Applied · 2019
Quantum simulation of the associative memory in Hopfield neural networks is an interesting crossover between quantum information and machine learning. The $q\phantom{\rule{0}{0ex}}u\phantom{\rule{0}{0ex}}a\phantom{\rule{0}{0ex}}n\phantom{\rule{0}{0ex}}t\phantom{\rule{0}{0ex}}u\phantom{\rule{0}{0ex}}m$ $s\phantom{\rule{0}{0ex}}t\phantom{\rule{0}{0ex}}o\phantom{\rule{0}{0ex}}c\phantom{\rule{0}{0ex}}h\phantom{\rule{0}{0ex}}a\phantom{\rule{0}{0ex}}s\phantom{\rule{0}{0ex}}t\phantom{\rule{0}{0ex}}i\phantom{\rule{0}{0ex}}c$ $w\phantom{\rule{0}{0ex}}a\phantom{\rule{0}{0ex}}l\phantom{\rule{0}{0ex}}k$ has been proposed for such simulations, yet not realized experimentally. The authors successfully map this scheme to a three-dimensional photonic chip, and achieve quantum stochastic walk evolution. A good match rate between the experimental quantum scheme and the expected result for a Hopfield neural net is attained. This proof of principle, combined with the scalability of low-loss integrated chips and straightforward Hamiltonian engineering, is a primary step toward photonic artificial intelligence.