Physical Neural Networks with Self-Learning Capabilities
arXiv:2408.05464 · doi:10.1007/s11433-024-2403-x
Abstract
Physical neural networks are artificial neural networks that mimic synapses and neurons using physical systems or materials. These networks harness the distinctive characteristics of physical systems to carry out computations effectively, potentially surpassing the constraints of conventional digital neural networks. A recent advancement known as ``physical self-learning'' aims to achieve learning through intrinsic physical processes rather than relying on external computations. This article offers a comprehensive review of the progress made in implementing physical self-learning across various physical systems. Prevailing learning strategies are discussed that contribute to the realization of physical self-learning. Despite challenges in understanding fundamental mechanism of learning, this work highlights the progress towards constructing intelligent hardware from the ground up, incorporating embedded self-organizing and self-adaptive dynamics in physical systems.
22 pages, 8 figures
References in corpus (50)
- Circuit Quantum Electrodynamics: Coherent Coupling of a Single Photon to a Cooper Pair Box
- Deep Learning with Coherent Nanophotonic Circuits
- Cavity quantum electrodynamics for superconducting electrical circuits: an architecture for quantum computation
- Training and Operation of an Integrated Neuromorphic Network Based on Metal-Oxide Memristors
- Recent Advances in Physical Reservoir Computing: A Review
- Photonics for artificial intelligence and neuromorphic computing
- Coupling Superconducting Qubits via a Cavity Bus
- Emergent complex neural dynamics
- Magnetization switching by spin-orbit torque in an antiferromagnet/ferromagnet bilayer system
- A high-bias, low-variance introduction to Machine Learning for physicists
- Deep physical neural networks enabled by a backpropagation algorithm for arbitrary physical systems
- Single-chip photonic deep neural network for instantaneous image classification
- Magnetic skyrmion artificial synapse for neuromorphic computing
- Vowel recognition with four coupled spin-torque nano-oscillators
- Training of photonic neural networks through in situ backpropagation
- Discovering physical concepts with neural networks
- Physically-informed artificial neural networks for atomistic modeling of materials
- Experimentally realized in situ backpropagation for deep learning in nanophotonic neural networks
- Memory formation in matter
- Magnetic skyrmion-based synaptic devices
- Challenges in materials and devices for Resistive-Switching-based Neuromorphic Computing
- Magnetic texture based magnonics
- Unsupervised Learning by Competing Hidden Units
- Reservoir Computing with Random Skyrmion Textures
- Emergence of Winner-takes-all Connectivity Paths in Random Nanowire Networks
- Quantum neuromorphic computing
- Learning without neurons in physical systems
- Reservoir Computing on Spin-Torque Oscillator Array
- Directed aging, memory and Nature's greed
- Demonstration of Decentralized, Physics-Driven Learning
- Quantum materials for energy-efficient neuromorphic computing
- Supervised learning in a mechanical system
- A memristive deep belief neural network based on silicon synapses
- Physics-informed neural networks for quantum control
- Discerning Incompressible and Compressible Phases of Cold Atoms in Optical Lattices
- Direct observation of a dynamical glass transition in a nanomagnetic artificial Hopfield network
- An atomic Boltzmann machine capable of on-chip learning
- Reservoir Computing with Spin Waves in Skyrmion Crystal
- Reconfigurable Reservoir Computing in a Magnetic Metamaterial
- Short-Term Plasticity and Long-Term Potentiation in Magnetic Tunnel Junctions: Towards Volatile Synapses
- Desynchronous Learning in a Physics-Driven Learning Network
- Signatures of associative memory behavior in a multi-mode spin-boson model
- Probabilistic Circuits for Autonomous Learning: A simulation study
- Stochastic single flux quantum neuromorphic computing using magnetically tunable Josephson junctions
- Predicting nonlinear optical scattering with physics-driven neural networks
- Antiferromagnet-Based Neuromorphics Using Dynamics of Topological Charges
- Quantum associative memory with a single driven-dissipative nonlinear oscillator
- Anticipative Tracking with the Short-Term Synaptic Plasticity of Spintronic Devices
- Stochastic spin-orbit-torque device as the STDP synapse for spiking neural networks
- Magnetic order in nanoscale gyroid networks