Error-aware probabilistic training for memristive neural networks
Jinchang Liu, Jian Lu, Shuangzhu Tang, Ruixi Zhou, Huiqin Ma, Bo Lyu, Yang Tian, Tuo Shi, Qi Liu · 2025
Read Error-aware probabilistic training for memristive neural networks by Jinchang Liu, Jian Lu, Shuangzhu Tang, Ruixi Zhou, Huiqin Ma, Bo Lyu, Yang Tian, Tuo Shi, Qi Liu (2025) 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/pmc-12749908.txt.
Description
Jinchang Liu and co-authors present a new way to train neural networks on memristor chips in 2025. These chips compute inside memory itself, which saves energy, but their random behavior clashes with the exact weight updates that standard training expects.
Their method, called EaPU, updates weights with a set chance drawn from the device's own writing noise. It needs far fewer updates than standard training, with little loss in accuracy. Tests on a real chip covered image cleaning and image sharpening tasks. The method improved accuracy by over sixty percent. It also cut training energy sharply compared with older chip-based training and with graphics cards.
Catalog record
| CREATOR | Jinchang Liu, Jian Lu, Shuangzhu Tang, Ruixi Zhou, Huiqin Ma, Bo Lyu, Yang Tian, Tuo Shi, Qi Liu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 55 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |