Contrastive learning in tunable dynamical systems
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866912986391642112 |
|---|---|
| author | Stern, Menachem Frim, Adam G. Candás, Raúl Liu, Andrea J. Balasubramanian, Vijay |
| author_facet | Stern, Menachem Frim, Adam G. Candás, Raúl Liu, Andrea J. Balasubramanian, Vijay |
| contents | We generalize the theory of supervised contrastive learning, previously applied to physical systems at equilibrium or steady state, to systems following any dynamics described by coupled ordinary differential equations. We show that if physical dynamics break time reversal symmetry, gradient descent on a cost function embodying the desired behavior cannot be achieved with a scalable process, even in principle. We therefore introduce Probably Approximately Right (PAR) learning processes, composed of a local contrastive learning rule and a scalable supervision protocol. We show that approximate, local supervision with forward propagation of the error signal can be used to successfully train several tunable models of physical dynamics inspired by examples in biological and machine learning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_26969 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Contrastive learning in tunable dynamical systems Stern, Menachem Frim, Adam G. Candás, Raúl Liu, Andrea J. Balasubramanian, Vijay Disordered Systems and Neural Networks Soft Condensed Matter Statistical Mechanics We generalize the theory of supervised contrastive learning, previously applied to physical systems at equilibrium or steady state, to systems following any dynamics described by coupled ordinary differential equations. We show that if physical dynamics break time reversal symmetry, gradient descent on a cost function embodying the desired behavior cannot be achieved with a scalable process, even in principle. We therefore introduce Probably Approximately Right (PAR) learning processes, composed of a local contrastive learning rule and a scalable supervision protocol. We show that approximate, local supervision with forward propagation of the error signal can be used to successfully train several tunable models of physical dynamics inspired by examples in biological and machine learning. |
| title | Contrastive learning in tunable dynamical systems |
| topic | Disordered Systems and Neural Networks Soft Condensed Matter Statistical Mechanics |
| url | https://arxiv.org/abs/2603.26969 |