Learned multi-stability in mechanical networks

Abstract

We contrast the distinct frameworks of materials design and physical learning in creating elastic networks with desired stable states. In design, the desired states are specified in advance and material parameters can be optimized on a computer with this knowledge. In learning, the material physically experiences the desired stable states in sequence, changing the material so as to stabilize each additional state. We show that while designed states are stable in networks of linear Hookean springs, sequential learning requires specific non-linear elasticity. We find that such non-linearity stabilizes states in which strain is zero in some springs and large in others, thus playing the role of Bayesian priors used in sparse statistical regression. Our model shows how specific material properties allow continuous learning of new functions through deployment of the material itself.

References in corpus (11)

• Overcoming catastrophic forgetting in neural networks
• Origami Multistabilty: From Single Vertices to Metasheets
• Modeling complex systems with adaptive networks
• Initial rigid response and softening transition of highly stretchable kirigami sheet materials
• Directed aging, memory and Nature's greed
• Multistable Kirigami for Tunable Architected Materials
• The limits of multifunctionality in tunable networks
• Localization from Incomplete Noisy Distance Measurements
• Supervised learning in a mechanical system
• Conformational Control of Mechanical Networks
• Minimal descriptions of cyclic memories

Cited by in corpus (9)

• Demonstration of Decentralized, Physics-Driven Learning
• Desynchronous Learning in a Physics-Driven Learning Network
• The Physical Effects of Learning
• Learning to self-fold at a bifurcation
• Drive-specific adaptation in disordered mechanical networks of bistable springs
• Memory capacity of adaptive flow networks
• Kinetic theory of decentralized learning for smart active matter
• Physical learning beyond the quasistatic limit
• Harnessing intuitive local evolution rules for physical learning