Emergence of biased errors in imperfect photonic circuits

Abstract

We study the impact of experimental imperfections in integrated photonic circuits. We discuss the emergence of a moderate biased error in path encoding, and investigate its correlation with properties of the optical paths. Our analysis connects and deepens previous studies in this direction, revealing potential issues for high-precision tests and optical implementations of machine learning.

14 pages, 10 figures, code available at https://zenodo.org/record/4924481

References in corpus (20)

• Photonics for artificial intelligence and neuromorphic computing
• 11 TeraFLOPs per second photonic convolutional accelerator for deep learning optical neural networks
• Integrated Photonic Quantum Technologies
• Quantum circuits with many photons on a programmable nanophotonic chip
• Reprogrammable Electro-Optic Nonlinear Activation Functions for Optical Neural Networks
• Experimental Scattershot Boson Sampling
• Design of optical neural networks with component imprecisions
• Experimental quantum speed-up in reinforcement learning agents
• Hardware error correction for programmable photonics
• A 12-mode Universal Photonic Processor for Quantum Information Processing
• Efficient training and design of photonic neural network through neuroevolution
• Optimal design of error-tolerant reprogrammable multiport interferometers
• Neural-Network Heuristics for Adaptive Bayesian Quantum Estimation
• Further Compactifying Linear Optical Unitaries
• Robust calibration of multiparameter sensors via machine learning at the single-photon level
• Benchmarking integrated photonic architectures
• Noise in BosonSampling and the threshold of efficient classical simulatability
• Photonic architecture for reinforcement learning
• Hybrid spatiotemporal architectures for universal linear optics
• Towards probing for hypercomplex quantum mechanics in a waveguide interferometer