Theory of neuromorphic computing by waves: machine learning by rogue waves, dispersive shocks, and solitons

Abstract

We study artificial neural networks with nonlinear waves as a computing reservoir. We discuss universality and the conditions to learn a dataset in terms of output channels and nonlinearity. A feed-forward three-layer model, with an encoding input layer, a wave layer, and a decoding readout, behaves as a conventional neural network in approximating mathematical functions, real-world datasets, and universal Boolean gates. The rank of the transmission matrix has a fundamental role in assessing the learning abilities of the wave. For a given set of training points, a threshold nonlinearity for universal interpolation exists. When considering the nonlinear Schroedinger equation, the use of highly nonlinear regimes implies that solitons, rogue, and shock waves do have a leading role in training and computing. Our results may enable the realization of novel machine learning devices by using diverse physical systems, as nonlinear optics, hydrodynamics, polaritonics, and Bose-Einstein condensates. The application of these concepts to photonics opens the way to a large class of accelerators and new computational paradigms. In complex wave systems, as multimodal fibers, integrated optical circuits, random, topological devices, and metasurfaces, nonlinear waves can be employed to perform computation and solve complex combinatorial optimization.

arXiv comment5 pages, 4 figures

References in corpus (17)

• Deep Learning in Neural Networks: An Overview
• Deep Learning with Coherent Nanophotonic Circuits
• All-Optical Machine Learning Using Diffractive Deep Neural Networks
• Stochastic Configuration Networks: Fundamentals and Algorithms
• All-optical Reservoir Computing
• Reinforcement Learning in a large scale photonic Recurrent Neural Network
• Wave Physics as an Analog Recurrent Neural Network
• Large-scale photonic Ising machine by spatial light modulation
• Heuristic Recurrent Algorithms for Photonic Ising Machines
• Optical Reservoir Computing using multiple light scattering for chaotic systems prediction
• Polaritonic neuromorphic computing outperforms linear classifiers
• Noise-enhanced spatial-photonic Ising machine
• Neuromorphic computing in Ginzburg-Landau polariton lattice systems
• Topological Control of Extreme Waves
• Leveraging Chaos for Wave-Based Analog Computation: Demonstration with Indoor Wireless Communication Signals
• From modulational instability to focusing dam breaks in water waves
• Universal linear optical operations on discrete phase-coherent spatial modes

Cited by in corpus (38)

• Nanoscale neural network using non-linear spin-wave interference
• Large-Scale Optical Reservoir Computing for Spatiotemporal Chaotic Systems Prediction
• Scalable Optical Learning Operator
• Opportunities in Quantum Reservoir Computing and Extreme Learning Machines
• Deep learning for dynamic modeling and coded information storage of vector-soliton pulsations in mode-locked fiber lasers
• Photonic extreme learning machine by free-space optical propagation
• Predicting the dynamic process and model parameters of the vector optical solitons in birefringent fibers via the modified PINN
• Pattern recognition with a magnon-scattering reservoir
• Scalable spin-glass optical simulator
• Photonic Extreme Learning Machine based on frequency multiplexing
• Neural Schrödinger Equation:Physical Law as Neural Network
• Wave-based extreme deep learning based on non-linear time-Floquet entanglement
• Self-learning Machines based on Hamiltonian Echo Backpropagation
• Efficient estimation of trainability for variational quantum circuits
• Mesoscopic ultrafast nonlinear optics -- The emergence of multimode quantum non-Gaussian physics
• Robust and Efficient Single-Pixel Image Classificationwith Nonlinear Optics
• A deep learning method for solving high-order nonlinear soliton equation
• Programmable metasurfaces for future photonic artificial intelligence
• Short-wavelength Reverberant Wave Systems for Physical Realization of Reservoir Computing
• Nonlinear Inference Capacity of Fiber-Optical Extreme Learning Machines
• Principles and Metrics of Extreme Learning Machines Using a Highly Nonlinear Fiber
• Photonic kernel machine learning for ultrafast spectral analysis
• Reservoir Computing with Superconducting Electronics
• Nonlinear Schrodinger Equation Solitons on Quantum Droplets
• Optical neuromorphic computing via temporal up-sampling and trainable encoding on a telecom device platform
• Predicting high-dimensional heterogeneous time series employing generalized local states
• Quantum-tunnelling deep neural network for optical illusion recognition
• Arbitrary control over multimode wave propagation for machine learning
• Limits of nonlinear and dispersive fiber propagation for an optical fiber-based extreme learning machine
• Thermodynamic limit in learning period three
• Optimization of a Hydrodynamic Computational Reservoir through Evolution
• Unsupervised Reservoir Computing for Solving Ordinary Differential Equations
• Soliton, Breather and Rogue Wave Solutions for Solving the Nonlinear Schrödinger Equation Using a Deep Learning Method with Physical Constraints
• Variational quantum algorithm for Gaussian discrete solitons and their boson sampling
• A universal approximation theorem for nonlinear resistive networks
• Gaussian boson sampling and multi-particle event optimization by machine learning in the quantum phase space
• Machine learning via relativity-inspired quantum dynamics
• Shannon Entropy Helps Optimize the Performance of a Frequency-Multiplexed Extreme Learning Machine