Repair Brain Damage: Real-Numbered Error Correction Code for Neural Network

Ziqing Li; Myung Cho; Qiutong Jin; Weiyu Xu

doi:10.48550/arxiv.2602.00076

Back

Repair Brain Damage: Real-Numbered Error Correction Code for Neural Network

Preprint

Open access

Repair Brain Damage: Real-Numbered Error Correction Code for Neural Network

Ziqing Li, Myung Cho, Qiutong Jin and Weiyu Xu

ArXiv.org

Cornell University

01/21/2026

DOI: 10.48550/arxiv.2602.00076

Files and links (1)

url

https://doi.org/10.48550/arxiv.2602.00076View

Preprint (Author's original)This preprint has not been evaluated by subject experts through peer review. Preprints may undergo extensive changes and/or become peer-reviewed journal articles. Open Access

Abstract

We consider a neural network (NN) that may experience memory faults and computational errors. In this paper, we propose a novel real-number-based error correction code (ECC) capable of detecting and correcting both memory errors and computational errors. The proposed approach introduces structures in the form of real-number-based linear constraints on the NN weights to enable error detection and correction, without sacrificing classification performance or increasing the number of real-valued NN parameters.

Computer Science - Learning

Computer Science - Neural and Evolutionary Computing

Details

Title: Subtitle: Repair Brain Damage: Real-Numbered Error Correction Code for Neural Network
Creators: Ziqing Li - University of Iowa
Myung Cho - California State University, Northridge
Qiutong Jin - University of California, Berkeley
Weiyu Xu - University of Iowa
Resource Type: Preprint
Publication Details: ArXiv.org
DOI: 10.48550/arxiv.2602.00076
ISSN: 2331-8422
Publisher: Cornell University; Ithaca, New York
Language: English
Date posted: 01/21/2026
Academic Unit: Electrical and Computer Engineering
Record Identifier: 9985139479202771

Metrics

2 Record Views