Backpropagation through nonlinear units for the all-optical training of neural networks
Photonics Research Optical Society of America 9:3 (2021) B71-B80
Abstract:
We propose a practical scheme for end-to-end optical backpropagation in neural networks. Using saturable absorption for the nonlinear units, we find that the backward-propagating gradients required to train the network can be approximated in a surprisingly simple pump-probe scheme that requires only simple passive optical elements. Simulations show that, with readily obtainable optical depths, our approach can achieve equivalent performance to state-of-the-art computational networks on image classification benchmarks, even in deep networks with multiple sequential gradient approximation. With backpropagation through nonlinear units being an outstanding challenge to the field, this work provides a feasible path toward truly all-optical neural networks.Reinforcement learning enhanced quantum-inspired algorithm for combinatorial optimization
Machine Learning: Science and Technology IOP Publishing 2:2 (2020) 025009
Abstract:
Quantum hardware and quantum-inspired algorithms are becoming increasingly popular for combinatorial optimization. However, these algorithms may require careful hyperparameter tuning for each problem instance. We use a reinforcement learning agent in conjunction with a quantum-inspired algorithm to solve the Ising energy minimization problem, which is equivalent to the Maximum Cut problem. The agent controls the algorithm by tuning one of its parameters with the goal of improving recently seen solutions. We propose a new Rescaled Ranked Reward (R3) method that enables a stable single-player version of self-play training and helps the agent escape local optima. The training on any problem instance can be accelerated by applying transfer learning from an agent trained on randomly generated problems. Our approach allows sampling high quality solutions to the Ising problem with high probability and outperforms both baseline heuristics and a black-box hyperparameter optimization approach.Quantum-enhanced interferometry with large heralded photon-number states
NPJ QUANTUM INFORMATION 6:1 (2020) ARTN 89
Abstract:
© 2020, The Author(s). Quantum phenomena such as entanglement can improve fundamental limits on the sensitivity of a measurement probe. In optical interferometry, a probe consisting of N entangled photons provides up to a N enhancement in phase sensitivity compared to a classical probe of the same energy. Here, we employ high-gain parametric down-conversion sources and photon-number-resolving detectors to perform interferometry with heralded quantum probes of sizes up to N = 8 (i.e. measuring up to 16-photon coincidences). Our probes are created by injecting heralded photon-number states into an interferometer, and in principle provide quantum-enhanced phase sensitivity even in the presence of significant optical loss. Our work paves the way toward quantum-enhanced interferometry using large entangled photonic states.Fully reconfigurable coherent optical vector–matrix multiplication
Optics Letters Optical Society of America 45:20 (2020) 5752-5755
Abstract:
Optics is a promising platform in which to help realize the next generation of fast, parallel, and energy-efficient computation. We demonstrate a reconfigurable free-space optical multiplier that is capable of over 3000 computations in parallel, using spatial light modulators with a pixel resolution of only 340×340. This enables vector–matrix multiplication and parallel vector–vector multiplication with vector size of up to 56. Our design is, to the best of our knowledge, the first to simultaneously support optical implementation of reconfigurable, large-sized, and real-valued linear algebraic operations. Such an optical multiplier can serve as a building block of special-purpose optical processors such as optical neural networks and optical Ising machines.Comprehensive model and performance optimization of phase-only spatial light modulators
Measurement Science and Technology IOP Publishing 31:12 (2020) 125202