Learning Dynamics in Memristor-Based Equilibrium Propagation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866910024315437056 |
|---|---|
| author | Döll, Michael Müller, Andreas Ulmann, Bernd |
| author_facet | Döll, Michael Müller, Andreas Ulmann, Bernd |
| contents | Memristor-based in-memory computing has emerged as a promising paradigm to overcome the constraints of the von Neumann bottleneck and the memory wall by enabling fully parallelisable and energy-efficient vector-matrix multiplications. We investigate the effect of nonlinear, memristor-driven weight updates on the convergence behaviour of neural networks trained with equilibrium propagation (EqProp). Six memristor models were characterised by their voltage-current hysteresis and integrated into the EBANA framework for evaluation on two benchmark classification tasks. EqProp can achieve robust convergence under nonlinear weight updates, provided that memristors exhibit a sufficiently wide resistance range of at least an order of magnitude. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_12428 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Learning Dynamics in Memristor-Based Equilibrium Propagation Döll, Michael Müller, Andreas Ulmann, Bernd Machine Learning Emerging Technologies Neural and Evolutionary Computing Memristor-based in-memory computing has emerged as a promising paradigm to overcome the constraints of the von Neumann bottleneck and the memory wall by enabling fully parallelisable and energy-efficient vector-matrix multiplications. We investigate the effect of nonlinear, memristor-driven weight updates on the convergence behaviour of neural networks trained with equilibrium propagation (EqProp). Six memristor models were characterised by their voltage-current hysteresis and integrated into the EBANA framework for evaluation on two benchmark classification tasks. EqProp can achieve robust convergence under nonlinear weight updates, provided that memristors exhibit a sufficiently wide resistance range of at least an order of magnitude. |
| title | Learning Dynamics in Memristor-Based Equilibrium Propagation |
| topic | Machine Learning Emerging Technologies Neural and Evolutionary Computing |
| url | https://arxiv.org/abs/2512.12428 |