Learning without neurons in physical systems

Abstract

Learning is traditionally studied in biological or computational systems. The power of learning frameworks in solving hard inverse-problems provides an appealing case for the development of `physical learning' in which physical systems adopt desirable properties on their own without computational design. It was recently realized that large classes of physical systems can physically learn through local learning rules, autonomously adapting their parameters in response to observed examples of use. We review recent work in the emerging field of physical learning, describing theoretical and experimental advances in areas ranging from molecular self-assembly to flow networks and mechanical materials. Physical learning machines provide multiple practical advantages over computer designed ones, in particular by not requiring an accurate model of the system, and their ability to autonomously adapt to changing needs over time. As theoretical constructs, physical learning machines afford a novel perspective on how physical constraints modify abstract learning theory.

25 pages, 6 figures

References in corpus (9)

• Programming Curvature using Origami Tessellations
• Combinatorial Design of Textured Mechanical Metamaterials
• Multifarious Assembly Mixtures: Systems Allowing Retrieval of Diverse Stored Structures
• Evolution of sparsity and modularity in a model of protein allostery
• Desynchronous Learning in a Physics-Driven Learning Network
• Active liquid crystals powered by force-sensing DNA-motor clusters
• A deep learning theory for neural networks grounded in physics
• Signal Transmissibility in Marginal Granular Materials
• Physical learning beyond the quasistatic limit

Cited by in corpus (23)

• A review of active matter reviews
• Soft Metamaterials: Adaptation and Intelligence
• Machine Learning Without a Processor: Emergent Learning in a Nonlinear Electronic Metamaterial
• Mechanical memories in solids, from disorder to design
• The Physical Effects of Learning
• Training Coupled Phase Oscillators as a Neuromorphic Platform using Equilibrium Propagation
• Rigidity of Epithelial Tissues as a Double Optimization Problem
• Intelligent mechanical metamaterials towards learning static and dynamic behaviors
• Qudit Machine Learning
• Physical Neural Networks with Self-Learning Capabilities
• Evidence of Athermal Metastable Phase in a Halide Perovskite: Optically Tracked Thermal-Breach Memory
• Mutual Multilinearity of Nonequilibrium Network Currents
• Self-organization and memory in an disordered solid subject to random loading
• Building rigid networks with prestress and selective pruning
• Physical networks become what they learn
• A universal approximation theorem for nonlinear resistive networks
• Memory capacity of adaptive flow networks
• Shape Switching and Tunable Oscillations of Adaptive Droplets
• A self-learning magnetic Hopfield neural network with intrinsic gradient descent adaption
• Les Houches Lectures on Flow Networks in Biology
• Irregular Metamaterial Networks
• From Active to Odd to Smart Matter
• Harnessing intuitive local evolution rules for physical learning