Analog Physical Systems Can Exhibit Double Descent
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915632571744256 |
|---|---|
| author | Dillavou, Sam Rocks, Jason W Wycoff, Jacob F Liu, Andrea J Durian, Douglas J |
| author_facet | Dillavou, Sam Rocks, Jason W Wycoff, Jacob F Liu, Andrea J Durian, Douglas J |
| contents | An important component of the success of large AI models is double descent, in which networks avoid overfitting as they grow relative to the amount of training data, instead improving their performance on unseen data. Here we demonstrate double descent in a decentralized analog network of self-adjusting resistive elements. This system trains itself and performs tasks without a digital processor, offering potential gains in energy efficiency and speed -- but must endure component non-idealities. We find that standard training fails to yield double descent, but a modified protocol that accommodates this inherent imperfection succeeds. Our findings show that analog physical systems, if appropriately trained, can exhibit behaviors underlying the success of digital AI. Further, they suggest that biological systems might similarly benefit from over-parameterization. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_17825 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Analog Physical Systems Can Exhibit Double Descent Dillavou, Sam Rocks, Jason W Wycoff, Jacob F Liu, Andrea J Durian, Douglas J Disordered Systems and Neural Networks Machine Learning An important component of the success of large AI models is double descent, in which networks avoid overfitting as they grow relative to the amount of training data, instead improving their performance on unseen data. Here we demonstrate double descent in a decentralized analog network of self-adjusting resistive elements. This system trains itself and performs tasks without a digital processor, offering potential gains in energy efficiency and speed -- but must endure component non-idealities. We find that standard training fails to yield double descent, but a modified protocol that accommodates this inherent imperfection succeeds. Our findings show that analog physical systems, if appropriately trained, can exhibit behaviors underlying the success of digital AI. Further, they suggest that biological systems might similarly benefit from over-parameterization. |
| title | Analog Physical Systems Can Exhibit Double Descent |
| topic | Disordered Systems and Neural Networks Machine Learning |
| url | https://arxiv.org/abs/2511.17825 |