Searcharxiv⌕ Search

arXiv subjects

Dumindu Samaraweera

Publications and source records attributed to Dumindu Samaraweera.

9 recordsLinked to original sources

Detection-Efficient Watermarking for Vision--Language Agents via Risk-Constrained Strength Allocation

Vision--language agents generate descriptions, decisions, and instructions that may be reused by users, downstream agents, and automated workflows, creating a need for provenance mechanisms that support content authenticity and post-deployment auditing without unnecessarily degrading response quality. We present a risk-aware semantic watermarking framework that combines predictive uncertainty with a zero-shot multimodal relevance-risk proxy to allocate watermark strength during decoding. A single risk penalty $λ$ exposes a continuous operating frontier, allowing deployment-time selection of a detection--quality trade-off without model retraining. At a selected operating point, mean watermark strength is reduced by approximately 59.4\% relative to fixed-strength watermarking, while perplexity change falls from 0.3634 to 0.1689 and clean/paraphrased ROC-AUC reaches 0.9552/0.9459. A matched-strength analysis shows that this advantage reflects the allocation policy rather than lower average intervention: at comparable mean strength, risk-constrained allocation achieves substantially higher ROC-AUC than fixed, entropy-only, and criticality-only policies, identifying it as detection-efficient rather than quality-dominant. We further show that threshold calibration on finite samples produces substantial false-positive-rate drift, motivating calibration variance reporting as standard practice for auditable provenance.

cs.CR↗

Watermark Evidence under Open-Weight LLM Modification: Exact Bounds for Low-Rank Output Watermarks

Open-weight language models can continue generating after their released weights are modified, while a watermark verifier still scores outputs against the original base and marked models. We study how such modifications change the evidence available to an OpenStamp-style verifier. At each shared prefix, we represent the modification as a residual in the next-token logits. We derive an exact finite-vocabulary lower bound on the expected log-likelihood-ratio evidence under a residual-span constraint, along with a sharp span-only deviation bound. We also characterize the first-order evidence shift through a covariance term with an explicit second-order remainder, and determine the residual span needed to reach a target evidence level. For sequence-level detection, we derive an exact conditional exponential-moment bound and combine it with a supermartingale argument and a distribution-free order-statistic tolerance bound. This yields a high-confidence upper bound on miss probability after attack selection on calibration data and evaluation on independent certification continuations. In experiments with OpenStamp on Llama-2-7B and Mistral-7B-v0.3, INT8 quantization and 30% pruning retain test true-positive rates of 85.6-94.8%, while 500-step LoRA fine-tuning lowers them to 28.8% and 11.6%. The covariance prediction tracks the measured mean LoRA evidence drift on both models, with no remainder-bound violations across 160,000 test prefixes. Additionally, we observe zero false positives among 500 base-model continuations per model. Through these theoretical and empirical results, the current research provides a better understanding of watermark robustness to model-editing/tuning attacks.

cs.CR↗

The Privacy and Security Risks of Edge LLM Optimization

Edge deployment of Large Language Models requires optimization techniques that reduce memory, computation, and energy costs. However, preserving task performance does not necessarily preserve security. We argue that optimization is an information transformation that can alter representations, decision pathways, and trust boundaries, creating security drift and optimization-induced attack surfaces.

cs.CR↗

Securing LLMs in the Wild: Privacy and Security Challenges at the Edge

Large Language Models (LLMs) are rapidly moving from research settings into the wild, deployed on enterprise infrastructure, personal devices, and edge platforms. While cloud deployments offer scalable compute, concerns over data sovereignty, compliance, latency, and third-party dependence are driving organizations toward edge and on-premise LLMs. This shift introduces new security and privacy challenges: limited compute and memory force aggressive optimizations, including quantization, pruning, model partitioning, and parameter-efficient adaptation, each of which can introduce vulnerabilities and reshape the threat landscape. We describe this tension as the Security-Efficiency Paradox, mechanisms that improve efficiency may weaken robustness, expose new attack surfaces, or increase privacy risks. We examine how compression can degrade safety alignment, how partitioned inference enables reconstruction attacks, and how continuous local adaptation may cause privacy leakage and model drift. To analyze these risks, we introduce a deployment-centric taxonomy organized around three architectural constraints: the Memory Wall, the Quadratic Wall, and the Compute Wall. We derive a unified constraint model that quantifies when unsafe optimizations become unavoidable, linking each wall to specific attack surfaces. Building on this model, we propose the Secure Operational Efficiency Score (SOES), a holistic metric balancing task accuracy, jailbreak resistance, and privacy against energy, memory, and latency, enabling practitioners to configure edge LLMs under real-world hardware limits. We further present a practical decision procedure and targeted mitigations for each optimization-induced vulnerability. Together, these contributions provide a co-designed framework for jointly evaluating security, privacy, and efficiency, laying a foundation for securing edge-native intelligent systems.

cs.CR↗

Building Trust in the Skies: A Knowledge-Grounded LLM-based Framework for Aviation Safety

The integration of Large Language Models (LLMs) into aviation safety decision-making represents a significant technological advancement, yet their standalone application poses critical risks due to inherent limitations such as factual inaccuracies, hallucination, and lack of verifiability. These challenges undermine the reliability required for safety-critical environments where errors can have catastrophic consequences. To address these challenges, this paper proposes a novel, end-to-end framework that synergistically combines LLMs and Knowledge Graphs (KGs) to enhance the trustworthiness of safety analytics. The framework introduces a dual-phase pipeline: it first employs LLMs to automate the construction and dynamic updating of an Aviation Safety Knowledge Graph (ASKG) from multimodal sources. It then leverages this curated KG within a Retrieval-Augmented Generation (RAG) architecture to ground, validate, and explain LLM-generated responses. The implemented system demonstrates improved accuracy and traceability over LLM-only approaches, effectively supporting complex querying and mitigating hallucination. Results confirm the framework's capability to deliver context-aware, verifiable safety insights, addressing the stringent reliability requirements of the aviation industry. Future work will focus on enhancing relationship extraction and integrating hybrid retrieval mechanisms.

cs.SE↗

Exploiting Meta-Learning-based Poisoning Attacks for Graph Link Prediction

Link prediction in graph data uses various algorithms and Graph Nerual Network (GNN) models to predict potential relationships between graph nodes. These techniques have found widespread use in numerous real-world applications, including recommendation systems, community/social networks, and biological structures. However, recent research has highlighted the vulnerability of GNN models to adversarial attacks, such as poisoning and evasion attacks. Addressing the vulnerability of GNN models is crucial to ensure stable and robust performance in GNN applications. Although many works have focused on enhancing the robustness of node classification on GNN models, the robustness of link prediction has received less attention. To bridge this gap, this article introduces an unweighted graph poisoning attack that leverages meta-learning with weighted scheme strategies to degrade the link prediction performance of GNNs. We conducted comprehensive experiments on diverse datasets across multiple link prediction applications to evaluate the proposed method and its parameters, comparing it with existing approaches under similar conditions. Our results demonstrate that our approach significantly reduces link prediction performance and consistently outperforms other state-of-the-art baselines.

cs.LG↗

Advancing Practical Homomorphic Encryption for Federated Learning: Theoretical Guarantees and Efficiency Optimizations

Federated Learning (FL) enables collaborative model training while preserving data privacy by keeping raw data locally stored on client devices, preventing access from other clients or the central server. However, recent studies reveal that sharing model gradients creates vulnerability to Model Inversion Attacks, particularly Deep Leakage from Gradients (DLG), which reconstructs private training data from shared gradients. While Homomorphic Encryption has been proposed as a promising defense mechanism to protect gradient privacy, fully encrypting all model gradients incurs high computational overhead. Selective encryption approaches aim to balance privacy protection with computational efficiency by encrypting only specific gradient components. However, the existing literature largely overlooks a theoretical exploration of the spectral behavior of encrypted versus unencrypted parameters, relying instead primarily on empirical evaluations. To address this gap, this paper presents a framework for theoretical analysis of the underlying principles of selective encryption as a defense against model inversion attacks. We then provide a comprehensive empirical study that identifies and quantifies the critical factors, such as model complexity, encryption ratios, and exposed gradients, that influence defense effectiveness. Our theoretical framework clarifies the relationship between gradient selection and privacy preservation, while our experimental evaluation demonstrates how these factors shape the robustness of defenses against model inversion attacks. Collectively, these contributions advance the understanding of selective encryption mechanisms and offer principled guidance for designing efficient, scalable, privacy-preserving federated learning systems.

cs.CR↗

Secure Distributed Learning for CAVs: Defending Against Gradient Leakage with Leveled Homomorphic Encryption

Federated Learning (FL) enables collaborative model training across distributed clients without sharing raw data, making it a promising approach for privacy-preserving machine learning in domains like Connected and Autonomous Vehicles (CAVs). However, recent studies have shown that exchanged model gradients remain susceptible to inference attacks such as Deep Leakage from Gradients (DLG), which can reconstruct private training data. While existing defenses like Differential Privacy (DP) and Secure Multi-Party Computation (SMPC) offer protection, they often compromise model accuracy. To that end, Homomorphic Encryption (HE) offers a promising alternative by enabling lossless computation directly on encrypted data, thereby preserving both privacy and model utility. However, HE introduces significant computational and communication overhead, which can hinder its practical adoption. To address this, we systematically evaluate various leveled HE schemes to identify the most suitable for FL in resource-constrained environments due to its ability to support fixed-depth computations without requiring costly bootstrapping. Our contributions in this paper include a comprehensive evaluation of HE schemes for real-world FL applications, a selective encryption strategy that targets only the most sensitive gradients to minimize computational overhead, and the development of a full HE-based FL pipeline that effectively mitigates DLG attacks while preserving model accuracy. We open-source our implementation to encourage reproducibility and facilitate adoption in safety-critical domains.

cs.CR↗

Cross-Model Transferability of Adversarial Patches in Real-time Segmentation for Autonomous Driving

Adversarial attacks pose a significant threat to deep learning models, particularly in safety-critical applications like healthcare and autonomous driving. Recently, patch based attacks have demonstrated effectiveness in real-time inference scenarios owing to their 'drag and drop' nature. Following this idea for Semantic Segmentation (SS), here we propose a novel Expectation Over Transformation (EOT) based adversarial patch attack that is more realistic for autonomous vehicles. To effectively train this attack we also propose a 'simplified' loss function that is easy to analyze and implement. Using this attack as our basis, we investigate whether adversarial patches once optimized on a specific SS model, can fool other models or architectures. We conduct a comprehensive cross-model transferability analysis of adversarial patches trained on SOTA Convolutional Neural Network (CNN) models such PIDNet-S, PIDNet-M and PIDNet-L, among others. Additionally, we also include the Segformer model to study transferability to Vision Transformers (ViTs). All of our analysis is conducted on the widely used Cityscapes dataset. Our study reveals key insights into how model architectures (CNN vs CNN or CNN vs. Transformer-based) influence attack susceptibility. In particular, we conclude that although the transferability (effectiveness) of attacks on unseen images of any dimension is really high, the attacks trained against one particular model are minimally effective on other models. And this was found to be true for both ViT and CNN based models. Additionally our results also indicate that for CNN-based models, the repercussions of patch attacks are local, unlike ViTs. Per-class analysis reveals that simple-classes like 'sky' suffer less misclassification than others. The code for the project is available at: https://github.com/p-shekhar/adversarial-patch-transferability

cs.CV↗