Adversarial Robustness in Machine Learning

Últimos artículos

• Why Safety Probes Catch Liars But Miss Fanatics

  Kristiyan Haralambiev · 30 de marzo de 2026

  Activation-based probes have emerged as a promising approach for detecting deceptively aligned AI systems by identifying internal conflict between true and stated goals. We identify a fundamental blind spot: probes fail on coherent misalignment - models that believe their harmful behavior is virtuou…

• Defending Against Knowledge Poisoning Attacks During Retrieval-Augmented Generation

  Kennedy Edemacu, Vinay M. Shashidhar, Micheal Tuape, Dan Abudu, Beakcheol Jang, Jong Wook Kim · 30 de marzo de 2026

  Retrieval-Augmented Generation (RAG) has emerged as a powerful approach to boost the capabilities of large language models (LLMs) by incorporating external, up-to-date knowledge sources. However, this introduces a potential vulnerability to knowledge poisoning attacks, where attackers can compromise…

• ProbGuard: Probabilistic Runtime Monitoring for LLM Agent Safety

  Haoyu Wang, Christopher M. Poskitt, Jiali Wei, Jun Sun · 30 de marzo de 2026

  Large Language Model (LLM) agents increasingly operate across domains such as robotics, virtual assistants, and web automation. However, their stochastic decision-making introduces safety risks that are difficult to anticipate during execution. Existing runtime monitoring frameworks, such as AgentSp…

• H-Node Attack and Defense in Large Language Models

  Eric Yocam, Varghese Vaidyan, Yong Wang · 30 de marzo de 2026

  We present H-Node Adversarial Noise Cancellation (H-Node ANC), a mechanistic framework that identifies, exploits, and defends hallucination representations in transformer-based large language models (LLMs) at the level of individual hidden-state dimensions. A logistic regression probe trained on las…

• Context-specific Credibility-aware Multimodal Fusion with Conditional Probabilistic Circuits

  Pranuthi Tenali, Sahil Sidheekh, Saurabh Mathur, Erik Blasch, Kristian Kersting, Sriraam Natarajan · 30 de marzo de 2026

  Multimodal fusion requires integrating information from multiple sources that may conflict depending on context. Existing fusion approaches typically rely on static assumptions about source reliability, limiting their ability to resolve conflicts when a modality becomes unreliable due to situational…

• UCAgent: An End-to-End Agent for Block-Level Functional Verification

  Junyue Wang, Zhicheng Yao, Yan Pi, Xiaolong Li, Fangyuan Song, Jinru Wang, Yunlong Xie, Sa Wang, Yungang Bao · 30 de marzo de 2026

  Functional verification remains a critical bottleneck in modern IC development cycles, accounting for approximately 70% of total development time in many projects. However, traditional methods, including constrained-random and formal verification, struggle to keep pace with the growing complexity of…

• A Compression Perspective on Simplicity Bias

  Tom Marty, Eric Elmoznino, Leo Gagnon, Tejas Kasetty, Mizu Nishikawa-Toomey, Sarthak Mittal, Guillaume Lajoie, Dhanya Sridhar · 30 de marzo de 2026

  Deep neural networks exhibit a simplicity bias, a well-documented tendency to favor simple functions over complex ones. In this work, we cast new light on this phenomenon through the lens of the Minimum Description Length principle, formalizing supervised learning as a problem of optimal two-part lo…

• Gap Safe Screening Rules for Fast Training of Robust Support Vector Machines under Feature Noise

  Tan-Hau Nguyen, Thu-Le Tran, Kien Trung Nguyen · 27 de marzo de 2026

  Robust Support Vector Machines (R-SVMs) address feature noise by adopting a worst-case robust formulation that explicitly incorporates uncertainty sets into training. While this robustness improves reliability, it also leads to increased computational cost. In this work, we develop safe sample scree…

• Generative Adversarial Perturbations with Cross-paradigm Transferability on Localized Crowd Counting

  Alabi Mehzabin Anisha, Guangjing Wang, Sriram Chellappan · 27 de marzo de 2026

  State-of-the-art crowd counting and localization are primarily modeled using two paradigms: density maps and point regression. Given the field's security ramifications, there is active interest in model robustness against adversarial attacks. Recent studies have demonstrated transferability across d…

• Knowledge-Guided Adversarial Training for Infrared Object Detection via Thermal Radiation Modeling

  Shiji Zhao, Shukun Xiong, Maoxun Yuan, Yao Huang, Ranjie Duan, Qing Guo, Jiansheng Chen, Haibin Duan, Xingxing Wei · 27 de marzo de 2026

  In complex environments, infrared object detection exhibits broad applicability and stability across diverse scenarios. However, infrared object detection is vulnerable to both common corruptions and adversarial examples, leading to potential security risks. To improve the robustness of infrared obj…

• On the Foundations of Trustworthy Artificial Intelligence

  TJ Dunham · 27 de marzo de 2026

  We prove that platform-deterministic inference is necessary and sufficient for trustworthy AI. We formalize this as the Determinism Thesis and introduce trust entropy to quantify the cost of non-determinism, proving that verification failure probability equals 1 - 2^{-H_T} exactly. We prove a …

• Sovereign AI at the Front Door of Care: A Physically Unidirectional Architecture for Secure Clinical Intelligence

  Vasu Srinivasan, Dhriti Vasu · 27 de marzo de 2026

  We present a Sovereign AI architecture for clinical triage in which all inference is performed on-device and inbound data is delivered via a physically unidirectional channel, implemented using receive-only broadcast infrastructure or certified hardware data diodes, with no return path to any extern…

• A Public Theory of Distillation Resistance via Constraint-Coupled Reasoning Architectures

  Peng Wei, Wesley Shu · 27 de marzo de 2026

  Knowledge distillation, model extraction, and behavior transfer have become central concerns in frontier AI. The main risk is not merely copying, but the possibility that useful capability can be transferred more cheaply than the governance structure that originally accompanied it. This paper presen…

• NERO-Net: A Neuroevolutionary Approach for the Design of Adversarially Robust CNNs

  In\^es Valentim, Nuno Antunes, Nuno Louren\c{c}o · 27 de marzo de 2026

  Neuroevolution automates the complex task of neural network design but often ignores the inherent adversarial fragility of evolved models which is a barrier to adoption in safety-critical scenarios. While robust training methods have received significant attention, the design of architectures exhibi…

• The Competence Shadow: Theory and Bounds of AI Assistance in Safety Engineering

  Umair Siddique · 27 de marzo de 2026

  As AI assistants become integrated into safety engineering workflows for Physical AI systems, a critical question emerges: does AI assistance improve safety analysis quality, or introduce systematic blind spots that surface only through post-deployment incidents? This paper develops a formal framewo…

• Beyond Content Safety: Real-Time Monitoring for Reasoning Vulnerabilities in Large Language Models

  Xunguang Wang, Yuguang Zhou, Qingyue Wang, Zongjie Li, Ruixuan Huang, Zhenlan Ji, Pingchuan Ma, Shuai Wang · 27 de marzo de 2026

  Large language models (LLMs) increasingly rely on explicit chain-of-thought (CoT) reasoning to solve complex tasks, yet the safety of the reasoning process itself remains largely unaddressed. Existing work on LLM safety focuses on content safety--detecting harmful, biased, or factually incorrect out…

• Mitigating Evasion Attacks in Fog Computing Resource Provisioning Through Proactive Hardening

  Younes Salmi, Hanna Bogucka · 27 de marzo de 2026

  This paper investigates the susceptibility to model integrity attacks that overload virtual machines assigned by the k-means algorithm used for resource provisioning in fog networks. The considered k-means algorithm runs two phases iteratively: offline clustering to form clusters of requested worklo…

• Decidable By Construction: Design-Time Verification for Trustworthy AI

  Houston Haynes · 27 de marzo de 2026

  A prevailing assumption in machine learning is that model correctness must be enforced after the fact. We observe that the properties determining whether an AI model is numerically stable, computationally correct, or consistent with a physical domain do not necessarily demand post hoc enforcement. T…

• Locket: Robust Feature-Locking Technique for Language Models

  Lipeng He, Vasisht Duddu, N. Asokan · 27 de marzo de 2026

  Chatbot service providers (e.g., OpenAI) rely on tiered subscription plans to generate revenue, offering black-box access to basic models for free users and advanced models to paying subscribers. However, this approach is unprofitable and inflexible for the users. A pay-to-unlock scheme for premium …

• AI Security in the Foundation Model Era: A Comprehensive Survey from a Unified Perspective

  Zhenyi Wang, Siyu Luan · 27 de marzo de 2026

  As machine learning (ML) systems expand in both scale and functionality, the security landscape has become increasingly complex, with a proliferation of attacks and defenses. However, existing studies largely treat these threats in isolation, lacking a coherent framework to expose their shared princ…

• DiffuGuard: How Intrinsic Safety is Lost and Found in Diffusion Large Language Models

  Zherui Li, Zheng Nie, Zhenhong Zhou, Yue Liu, Yitong Zhang, Yu Cheng, Qingsong Wen, Kun Wang, Yufei Guo, Jiaheng Zhang · 27 de marzo de 2026

  The rapid advancement of Diffusion Large Language Models (dLLMs) introduces unprecedented vulnerabilities that are fundamentally distinct from Autoregressive LLMs, stemming from their iterative and parallel generation mechanisms. In this paper, we conduct an in-depth analysis of dLLM vulnerabilities…

• SafeMath: Inference-time Safety improves Math Accuracy

  Sagnik Basu, Subhrajit Mitra, Aman Juneja, Somnath Banerjee, Rima Hazra, Animesh Mukherjee · 27 de marzo de 2026

  Recent research points toward LLMs being manipulated through adversarial and seemingly benign inputs, resulting in harmful, biased, or policy-violating outputs. In this paper, we study an underexplored issue concerning harmful and toxic mathematical word problems. We show that math questions, partic…

• Safe Reinforcement Learning with Preference-based Constraint Inference

  Chenglin Li, Guangchun Ruan, Hua Geng · 26 de marzo de 2026

  Safe reinforcement learning (RL) is a standard paradigm for safety-critical decision making. However, real-world safety constraints can be complex, subjective, and even hard to explicitly specify. Existing works on constraint inference rely on restrictive assumptions or extensive expert demonstratio…

• How Vulnerable Are Edge LLMs?

  Ao Ding, Hongzong Li, Zi Liang, Zhanpeng Shi, Shuxin Zhuang, Shiqin Tang, Rong Feng, Ping Lu · 26 de marzo de 2026

  Large language models (LLMs) are increasingly deployed on edge devices under strict computation and quantization constraints, yet their security implications remain unclear. We study query-based knowledge extraction from quantized edge-deployed LLMs under realistic query budgets and show that, altho…

• The Luna Bound Propagator for Formal Analysis of Neural Networks

  Henry LeCates, Haoze Wu · 26 de marzo de 2026

  The parameterized CROWN analysis, a.k.a., alpha-CROWN, has emerged as a practically successful bound propagation method for neural network verification. However, existing implementations of alpha-CROWN are limited to Python, which complicates integration into existing DNN verifiers and long-term pro…