Research

Paper

AI LLM March 20, 2026

Neural Uncertainty Principle: A Unified View of Adversarial Fragility and LLM Hallucination

Authors

Dong-Xiao Zhang, Hu Lou, Jun-Jie Zhang, Jun Zhu, Deyu Meng

Abstract

Adversarial vulnerability in vision and hallucination in large language models are conventionally viewed as separate problems, each addressed with modality-specific patches. This study first reveals that they share a common geometric origin: the input and its loss gradient are conjugate observables subject to an irreducible uncertainty bound. Formalizing a Neural Uncertainty Principle (NUP) under a loss-induced state, we find that in near-bound regimes, further compression must be accompanied by increased sensitivity dispersion (adversarial fragility), while weak prompt-gradient coupling leaves generation under-constrained (hallucination). Crucially, this bound is modulated by an input-gradient correlation channel, captured by a specifically designed single-backward probe. In vision, masking highly coupled components improves robustness without costly adversarial training; in language, the same prefill-stage probe detects hallucination risk before generating any answer tokens. NUP thus turns two seemingly separate failure taxonomies into a shared uncertainty-budget view and provides a principled lens for reliability analysis. Guided by this NUP theory, we propose ConjMask (masking high-contribution input components) and LogitReg (logit-side regularization) to improve robustness without adversarial training, and use the probe as a decoding-free risk signal for LLMs, enabling hallucination detection and prompt selection. NUP thus provides a unified, practical framework for diagnosing and mitigating boundary anomalies across perception and generation tasks.

Metadata

arXiv ID: 2603.19562
Provider: ARXIV
Primary Category: cs.LG
Published: 2026-03-20
Fetched: 2026-03-23 16:54

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.19562v1</id>\n    <title>Neural Uncertainty Principle: A Unified View of Adversarial Fragility and LLM Hallucination</title>\n    <updated>2026-03-20T02:07:10Z</updated>\n    <link href='https://arxiv.org/abs/2603.19562v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.19562v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Adversarial vulnerability in vision and hallucination in large language models are conventionally viewed as separate problems, each addressed with modality-specific patches. This study first reveals that they share a common geometric origin: the input and its loss gradient are conjugate observables subject to an irreducible uncertainty bound. Formalizing a Neural Uncertainty Principle (NUP) under a loss-induced state, we find that in near-bound regimes, further compression must be accompanied by increased sensitivity dispersion (adversarial fragility), while weak prompt-gradient coupling leaves generation under-constrained (hallucination). Crucially, this bound is modulated by an input-gradient correlation channel, captured by a specifically designed single-backward probe. In vision, masking highly coupled components improves robustness without costly adversarial training; in language, the same prefill-stage probe detects hallucination risk before generating any answer tokens. NUP thus turns two seemingly separate failure taxonomies into a shared uncertainty-budget view and provides a principled lens for reliability analysis. Guided by this NUP theory, we propose ConjMask (masking high-contribution input components) and LogitReg (logit-side regularization) to improve robustness without adversarial training, and use the probe as a decoding-free risk signal for LLMs, enabling hallucination detection and prompt selection. NUP thus provides a unified, practical framework for diagnosing and mitigating boundary anomalies across perception and generation tasks.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.LG'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.IT'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='physics.comp-ph'/>\n    <published>2026-03-20T02:07:10Z</published>\n    <arxiv:comment>16 pages,3 figures</arxiv:comment>\n    <arxiv:primary_category term='cs.LG'/>\n    <author>\n      <name>Dong-Xiao Zhang</name>\n    </author>\n    <author>\n      <name>Hu Lou</name>\n    </author>\n    <author>\n      <name>Jun-Jie Zhang</name>\n    </author>\n    <author>\n      <name>Jun Zhu</name>\n    </author>\n    <author>\n      <name>Deyu Meng</name>\n    </author>\n  </entry>"
}