Adversarial Robustness in Machine Learning

Neueste Paper

• Steering MoE LLMs via Expert (De)Activation

  Mohsen Fayyaz, Ali Modarressi, Hanieh Deilamsalehy, Franck Dernoncourt, Ryan Rossi, Trung Bui, Hinrich Sch\"utze, Nanyun Peng · 13. Februar 2026

  Mixture-of-Experts (MoE) in Large Language Models (LLMs) routes each token through a subset of specialized Feed-Forward Networks (FFN), known as experts. We present SteerMoE, a framework to steer MoE models by detecting and controlling behavior-associated experts. We detect key experts by comparing …

• Revis: Sparse Latent Steering to Mitigate Object Hallucination in Large Vision-Language Models

  Jialin Wu, Wei Shi, Han Shen, Peigui Qi, Kunsheng Tang, Zhicong Huang, Binghao Wang, Zhou Yang · 13. Februar 2026

  Despite the advanced capabilities of Large Vision-Language Models (LVLMs), they frequently suffer from object hallucination. One reason is that visual features and pretrained textual representations often become intertwined in the deeper network layers. To address this, we propose REVIS, a training-…

• Amortised and provably-robust simulation-based inference

  Ayush Bharti, Charita Dellaporta, Yuga Hikida, Fran\c{c}ois-Xavier Briol · 13. Februar 2026

  Complex simulator-based models are now routinely used to perform inference across the sciences and engineering, but existing inference methods are often unable to account for outliers and other extreme values in data which occur due to faulty measurement instruments or human error. In this paper, we…

• Sample-Free Safety Assessment of Neural Network Controllers via Taylor Methods

  Adam Evans, Roberto Armellin · 13. Februar 2026

  In recent years, artificial neural networks have been increasingly studied as feedback controllers for guidance problems. While effective in complex scenarios, they lack the verification guarantees found in classical guidance policies. Their black-box nature creates significant concerns regarding tr…

• Sparse Semantic Dimension as a Generalization Certificate for LLMs

  Dibyanayan Bandyopadhyay, Asif Ekbal · 13. Februar 2026

  Standard statistical learning theory predicts that Large Language Models (LLMs) should overfit because their parameter counts vastly exceed the number of training tokens. Yet, in practice, they generalize robustly. We propose that the effective capacity controlling generalization lies in the geometr…

• ANML: Attribution-Native Machine Learning with Guaranteed Robustness

  Oliver Zahn, Matt Beton, Simran Chana · 13. Februar 2026

  Frontier AI systems increasingly train on specialized expert data, from clinical records to proprietary research to curated datasets, yet current training pipelines treat all samples identically. A Nobel laureate's contribution receives the same weight as an unverified submission. We introduce ANML …

• Agentic AI for Cybersecurity: A Meta-Cognitive Architecture for Governable Autonomy

  Andrei Kojukhov, Arkady Bovshover · 13. Februar 2026

  Contemporary AI-driven cybersecurity systems are predominantly architected as model-centric detection and automation pipelines optimized for task-level performance metrics such as accuracy and response latency. While effective for bounded classification tasks, these architectures struggle to support…

• SafeNeuron: Neuron-Level Safety Alignment for Large Language Models

  Zhaoxin Wang, Jiaming Liang, Fengbin Zhu, Weixiang Zhao, Junfeng Fang, Jiayi Ji, Handing Wang, Tat-Seng Chua · 13. Februar 2026

  Large language models (LLMs) and multimodal LLMs are typically safety-aligned before release to prevent harmful content generation. However, recent studies show that safety behaviors are concentrated in a small subset of parameters, making alignment brittle and easily bypassed through neuron-level a…

• The Energy of Falsehood: Detecting Hallucinations via Diffusion Model Likelihoods

  Arpit Singh Gautam, Kailash Talreja, Saurabh Jha · 13. Februar 2026

  Large Language Models (LLMs) frequently hallucinate plausible but incorrect assertions, a vulnerability often missed by uncertainty metrics when models are confidently wrong. We propose DiffuTruth, an unsupervised framework that reconceptualizes fact verification via non equilibrium thermodynamics, …

• Brain Tumor Classifiers Under Attack: Robustness of ResNet Variants Against Transferable FGSM and PGD Attacks

  Ryan Deem, Garrett Goodman, Waqas Majeed, Md Abdullah Al Hafiz Khan, Michail S. Alexiou · 13. Februar 2026

  Adversarial robustness in deep learning models for brain tumor classification remains an underexplored yet critical challenge, particularly for clinical deployment scenarios involving MRI data. In this work, we investigate the susceptibility and resilience of several ResNet-based architectures, refe…

• Evaluating LLM Safety Under Repeated Inference via Accelerated Prompt Stress Testing

  Keita Broadwater · 13. Februar 2026

  Traditional benchmarks for large language models (LLMs) primarily assess safety risk through breadth-oriented evaluation across diverse tasks. However, real-world deployment exposes a different class of risk: operational failures arising from repeated inference on identical or near-identical prompts…

• SKATE, a Scalable Tournament Eval: Weaker LLMs differentiate between stronger ones using verifiable challenges

  Dewi S. W. Gould, Bruno Mlodozeniec, Samuel F. Brown · 13. Februar 2026

  Evaluating the capabilities and risks of foundation models is paramount, yet current methods demand extensive domain expertise, hindering their scalability as these models rapidly evolve. We introduce SKATE: a novel evaluation framework in which large language models (LLMs) compete by generating and…

• Capability-Oriented Training Induced Alignment Risk

  Yujun Zhou, Yue Huang, Han Bao, Kehan Guo, Zhenwen Liang, Pin-Yu Chen, Tian Gao, Werner Geyer, Nuno Moniz, Nitesh V Chawla, Xiangliang Zhang · 13. Februar 2026

  While most AI alignment research focuses on preventing models from generating explicitly harmful content, a more subtle risk is emerging: capability-oriented training induced exploitation. We investigate whether language models, when trained with reinforcement learning (RL) in environments with impl…

• Safety Beyond the Training Data: Robust Out-of-Distribution MPC via Conformalized System Level Synthesis

  Anutam Srinivasan, Antoine Leeman, Glen Chou · 13. Februar 2026

  We present a novel framework for robust out-of-distribution planning and control using conformal prediction (CP) and system level synthesis (SLS), addressing the challenge of ensuring safety and robustness when using learned dynamics models beyond the training data distribution. We first derive high…

• AgentLeak: A Full-Stack Benchmark for Privacy Leakage in Multi-Agent LLM Systems

  Faouzi El Yagoubi, Ranwa Al Mallah, Godwin Badu-Marfo · 13. Februar 2026

  Multi-agent Large Language Model (LLM) systems create privacy risks that current benchmarks cannot measure. When agents coordinate on tasks, sensitive data passes through inter-agent messages, shared memory, and tool arguments; pathways that output-only audits never inspect. We introduce AgentLeak, …

• TIP: Resisting Gradient Inversion via Targeted Interpretable Perturbation in Federated Learning

  Jianhua Wang, Yinlin Su · 13. Februar 2026

  Federated Learning (FL) facilitates collaborative model training while preserving data locality; however, the exchange of gradients renders the system vulnerable to Gradient Inversion Attacks (GIAs), allowing adversaries to reconstruct private training data with high fidelity. Existing defenses, suc…

• DeepSight: An All-in-One LM Safety Toolkit

  Bo Zhang, Jiaxuan Guo, Lijun Li, Dongrui Liu, Sujin Chen, Guanxu Chen, Zhijie Zheng, Qihao Lin, Lewen Yan, Chen Qian, Yijin Zhou, Yuyao Wu, Shaoxiong Guo, Tianyi Du, Jingyi Yang, Xuhao Hu, Ziqi Miao, Xiaoya Lu, Jing Shao, Xia Hu · 13. Februar 2026

  As the development of Large Models (LMs) progresses rapidly, their safety is also a priority. In current Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) safety workflow, evaluation, diagnosis, and alignment are often handled by separate tools. Specifically, safety evaluatio…

• Temporally Unified Adversarial Perturbations for Time Series Forecasting

  Ruixian Su, Yukun Bao, Xinze Zhang · 13. Februar 2026

  While deep learning models have achieved remarkable success in time series forecasting, their vulnerability to adversarial examples remains a critical security concern. However, existing attack methods in the forecasting field typically ignore the temporal consistency inherent in time series data, l…

• Detecting RLVR Training Data via Structural Convergence of Reasoning

  Hongbo Zhang, Yue Yang, Jianhao Yan, Guangsheng Bao, Yue Zhang, Yue Zhang · 13. Februar 2026

  Reinforcement learning with verifiable rewards (RLVR) is central to training modern reasoning models, but the undisclosed training data raises concerns about benchmark contamination. Unlike pretraining methods, which optimize models using token-level probabilities, RLVR fine-tunes models based on re…

• Zero-Sacrifice Persistent-Robustness Adversarial Defense for Pre-Trained Encoders

  Zhuxin Lei, Ziyuan Yang, Yi Zhang · 13. Februar 2026

  The widespread use of publicly available pre-trained encoders from self-supervised learning (SSL) has exposed a critical vulnerability: their susceptibility to downstream-agnostic adversarial examples (DAEs), which are crafted without knowledge of the downstream tasks but capable of misleading downs…

• Gradients Must Earn Their Influence: Unifying SFT with Generalized Entropic Objectives

  Zecheng Wang, Deyuan Liu, Chunshan Li, Yupeng Zhang, Zhengyun Zhao, Dianhui Chu, Bingning Wang, Dianbo Sui · 13. Februar 2026

  Standard negative log-likelihood (NLL) for Supervised Fine-Tuning (SFT) applies uniform token-level weighting. This rigidity creates a two-fold failure mode: (i) overemphasizing low-probability targets can amplify gradients on noisy supervision and disrupt robust priors, and (ii) uniform weighting p…

• Towards Reliable Machine Translation: Scaling LLMs for Critical Error Detection and Safety

  Muskaan Chopra, Lorenz Sparrenberg, Rafet Sifa · 13. Februar 2026

  Machine Translation (MT) plays a pivotal role in cross-lingual information access, public policy communication, and equitable knowledge dissemination. However, critical meaning errors, such as factual distortions, intent reversals, or biased translations, can undermine the reliability, fairness, and…

• In-the-Wild Model Organisms: Mitigating Undesirable Emergent Behaviors in Production LLM Post-Training via Data Attribution

  Frank Xiao, Santiago Aranguri · 12. Februar 2026

  We propose activation-based data attribution, a method that traces behavioral changes in post-trained language models to responsible training datapoints. By computing activation-difference vectors for both test prompts and preference pairs and ranking by cosine similarity, we identify datapoints tha…

• OmniVL-Guard: Towards Unified Vision-Language Forgery Detection and Grounding via Balanced RL

  Jinjie Shen, Jing Wu, Yaxiong Wang, Lechao Cheng, Shengeng Tang, Tianrui Hui, Nan Pu, Zhun Zhong · 12. Februar 2026

  Existing forgery detection methods are often limited to uni-modal or bi-modal settings, failing to handle the interleaved text, images, and videos prevalent in real-world misinformation. To bridge this gap, this paper targets to develop a unified framework for omnibus vision-language forgery detecti…

• Enhancing Inverse Reinforcement Learning through Encoding Dynamic Information in Reward Shaping

  Simon Sinong Zhan, Philip Wang, Qingyuan Wu, Yixuan Wang, Ruochen Jiao, Chao Huang, Qi Zhu · 12. Februar 2026

  In this paper, we aim to tackle the limitation of the Adversarial Inverse Reinforcement Learning (AIRL) method in stochastic environments where theoretical results cannot hold and performance is degraded. To address this issue, we propose a novel method which infuses the dynamics information into th…