Paper Detail

Maximal Brain Damage Without Data or Optimization: Disrupting Neural Networks via Sign-Bit Flips

Ido Galil, Moshe Kimhi, Ran El-Yaniv

huggingface Score 9.5

Published 2026-04-16 · First seen 2026-04-20

General AI

Abstract

Deep Neural Networks (DNNs) can be catastrophically disrupted by flipping only a handful of parameter bits. We introduce Deep Neural Lesion (DNL), a data-free and optimizationfree method that locates critical parameters, and an enhanced single-pass variant, 1P-DNL, that refines this selection with one forward and backward pass on random inputs. We show that this vulnerability spans multiple domains, including image classification, object detection, instance segmentation, and reasoning large language models. In image classification, flipping just two sign bits in ResNet-50 on ImageNet reduces accuracy by 99.8%. In object detection and instance segmentation, one or two sign flips in the backbone collapse COCO detection and mask AP for Mask R-CNN and YOLOv8-seg models. In language modeling, two sign flips into different experts reduce Qwen3-30B-A3B-Thinking from 78% to 0% accuracy. We also show that selectively protecting a small fraction of vulnerable sign bits provides a practical defense against such attacks.

Workflow Status

Review status
pending
Role
unreviewed
Read priority
soon
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

BibTeX

@misc{galil2026maximal,
  title = {Maximal Brain Damage Without Data or Optimization: Disrupting Neural Networks via Sign-Bit Flips},
  author = {Ido Galil and Moshe Kimhi and Ran El-Yaniv},
  year = {2026},
  abstract = {Deep Neural Networks (DNNs) can be catastrophically disrupted by flipping only a handful of parameter bits. We introduce Deep Neural Lesion (DNL), a data-free and optimizationfree method that locates critical parameters, and an enhanced single-pass variant, 1P-DNL, that refines this selection with one forward and backward pass on random inputs. We show that this vulnerability spans multiple domains, including image classification, object detection, instance segmentation, and reasoning large lang},
  url = {https://huggingface.co/papers/2502.07408},
  keywords = {Deep Neural Networks, parameter bits, catastrophic disruption, Deep Neural Lesion, 1P-DNL, sign bits, ResNet-50, ImageNet, object detection, instance segmentation, Mask R-CNN, YOLOv8-seg, language modeling, Qwen3-30B-A3B-Thinking, code available, huggingface daily},
  eprint = {2502.07408},
  archiveprefix = {arXiv},
}

Metadata

{}