How to Train Your Resistive Network: Generalized Equilibrium Propagation and Analytical Learning
J. Lin, A. Desai, F. Barrows, F. Caravelli · 2026
Read How to Train Your Resistive Network: Generalized Equilibrium Propagation and Analytical Learning by J. Lin, A. Desai, F. Barrows, F. Caravelli (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-03546.txt.
Description
Digital machine learning burns enormous energy, and analog resistor networks might one day do the same work cheaply. Jonathan Lin and 3 co-authors tackle the training problem for such hardware in a 2026 paper. Physical systems demand local learning rules, but the known ones, like equilibrium propagation, come with awkward requirements such as full readouts.
The authors derive exact gradients analytically from Kirchhoff's laws using a graph-theoretic framework. They also introduce Generalized Equilibrium Propagation, which unifies a broad family of Hebbian learning rules, including coupled learning. Simulations show resistor networks can train while reading only the output layer, with no replica network. Updating only a subset of resistances degrades performance surprisingly little.
Catalog record
| CREATOR | J. Lin, A. Desai, F. Barrows, F. Caravelli |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 87 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-03546