Searcharxiv⌕ Search

arXiv subjects

Mohan Gurusamy

Publications and source records attributed to Mohan Gurusamy.

At least 19 recordsLinked to original sources

The Truth Beneath the Tokens: Prompt-Space Recovery of Knowledge Edits in Language Models

Model editing provides an efficient way to update factual knowledge in transformer models. Recent approaches reverse edits by modifying, masking, or decomposing model weights, but make recovery dependent on weight-space interventions. We instead show that pre-edit knowledge can be recovered through input-space interventions alone, without modifying the edited model's weights. We introduce Memento, a prompt-space mechanism for auditing residual knowledge that learns a reusable recovery signal to steer an edited model toward its pre-edit response. Memento supports both continuous recovery prefixes and discrete textual prompts for standard text interfaces. Across three LLMs and five editing techniques, Memento recovers 80.07% of held-out edited facts, achieving a 9.58% relative improvement over baselines, suggesting that editing may suppress the facts without fully eliminating their recoverability. We further investigate whether prompt-space recovery extends beyond model editing. On pristine, unedited LLMs, Memento improves average factual recall by 53.9% relative to standard inference. These results demonstrate the potential of prompt-space recovery for auditing residual knowledge in edited models and investigating the gap between factual knowledge and its accessibility during inference. We will release our source code.

cs.CL↗

CLaRE-ty Amid Chaos: Quantifying Representational Entanglement to Predict Ripple Effects in LLM Editing

The static knowledge representations of large language models (LLMs) inevitably become outdated or incorrect over time. While model-editing techniques offer a promising solution by modifying a model's factual associations, they often produce unpredictable ripple effects, which are unintended behavioral changes that propagate even to the hidden space. In this work, we introduce CLaRE, a lightweight representation-level technique to identify where these ripple effects may occur. Unlike prior gradient-based methods, CLaRE quantifies entanglement between facts using forward activations from a single intermediate layer, avoiding costly backward passes. To enable systematic study, we prepare and analyse a corpus of 11,427 facts drawn from three existing datasets. Using CLaRE, we compute large-scale entanglement graphs of this corpus for multiple models, capturing how local edits propagate through representational space. These graphs enable stronger preservation sets for model editing, audit trails, efficient red-teaming, and scalable post-edit evaluation. In comparison to baselines, CLaRE achieves an average of 62.2% improvement in Spearman correlation with ripple effects while being 2.7x faster, and using 2.85x less peak GPU memory. Besides, CLaRE requires only a fraction of the storage needed by the baselines to compute and preserve fact representations. Our entanglement graphs and corpus are available at https://github.com/manitbaser/CLaRE.

cs.LG↗

NetGuise: Adversarial Traffic Generation for Early-Stage Classifier Evasion

Early-stage encrypted traffic classifiers make decisions from only the first few packets of a connection, enabling low-latency classification but raising robustness concerns under realistic traffic morphing. Existing adversarial traffic generation methods often assume complete flows, feature-level perturbations, or iterative runtime feedback, which do not match the constraints of early-stage execution. We present NetGuise, a target-conditioned traffic morphing framework that generates packet-level padding and non-negative delay actions while preserving packet direction, ordering, and packet count and respecting MTU, bandwidth, and latency constraints. We evaluate NetGuise for censorship evasion under two observation interfaces, i) a history-only causal interface for online packet-by-packet execution, ii) and a full-window oracle interface that provides an upper-bound characterization of early-stage morphability. On VANT, NetGuise achieves 33.2% mean targeted evasion under the causal interface versus 82.0% under the oracle interface; on UTMobileNet, the corresponding results are 5.6% and 81.8%. NetGuise outperforms the strongest causal baseline, BAP, on both datasets and, on VANT, reduces bandwidth overhead from 19.80% to 3.02% and generation latency from 37.42 ms/flow to 1.16 ms/flow. Under the oracle interface, it achieves up to 85.7% targeted evasion, retains 43.5% after four rounds of adaptive censor retraining, and maintains 48% under zero-shot cross-environment transfer. Our results show that early-stage classifiers can remain morphable despite high clean accuracy and that robustness evaluation should account for targeted morphability, observation constraints, defender adaptation, and operational network conditions.

cs.NI↗

Before Retrieval Begins: Safety and Tool-Use Decisions in LLM Agents

Large language models increasingly act as agents that call external tools, yet most tool-safety research examines only what follows a tool call. The prior decision of whether to call the tool remains largely unexamined and can influence downstream safety. Across the seven open-weight models, the unweighted model-average harm rate after admitted harmful requests is 74.5%. To identify what governs this decision, we apply a factorial intervention to the input request. Within our generated construction, neutralizing the requested-operation component produces the largest admission contrast (19 pp). A linear direction estimated from this contrast influences held-out admission in four of seven models; effects on operation-neutralized requests in some models show that it is not an operation-specific mechanism. We then test whether written refusal is behaviorally and representationally coupled to the same decision. Written refusal accompanies 33.6% of admitted harmful calls. Plain-text refusal training reduces unsafe admission in models that retain benign tool use, showing that response-level safety training can transfer to the call decision. Together, these results establish a partial, model-dependent relation between written refusal and tool admission. They motivate direct evaluation and supervision of tool admission rather than treating text refusal as its proxy.

cs.CL↗

Disentangling Language Effects on Tool Admission and Post-Retrieval Safety

LLM agents increasingly use retrieval tools across languages, yet their safety evaluation remains English-centric. It remains unclear whether language affects tool admission, post-retrieval generation, or both. We independently vary request, retrieved-page, and required response languages in a controlled URL-retrieval setting across ten models and seven languages. We identify Safety Loss through Interlingual Processing (SLIP). Non-English harmful requests increase mean tool admission from 46.9% to 67.4%, with a positive mean change in every model. This pattern also persists in an autonomous retrieval setup across all four models tested in this setting. Requiring answers in the page language increases pooled harmfulness on matched tool calls from 50.3% to 57.4% across four closed-weight models, mainly in Chinese and Hindi. Translating pages alone produces no consistent harm increase in eight models with comparable inference settings. Automated audits show that a non-harmful response may be a genuine safety refusal, a deflection citing language difficulties, or an unusable answer. Harmfulness labels alone cannot distinguish these outcomes. Multilingual safety evaluations should therefore measure tool admission and response harmfulness separately and distinguish refusal from capability failure.

cs.CL↗

The Fragility of Trigger-Tag Mechanisms for Misuse Detection in Open-Weight LLMs

Open-weight language models can be downloaded, modified, and deployed beyond their developers' control, limiting the effectiveness of centrally enforced safeguards. Recent work has therefore proposed \emph{trigger-tag} mechanisms that produce a detectable signal when a model is used under a target condition, such as generating phishing contents. Although these mechanisms borrow from established techniques, their use for conditional misuse detection in open-weight LLMs is relatively new. Therefore, existing research works have not systematically studied the robustness of trigger-tag mechanisms under adversarial attacks. To close this gap, (i)~we formalize trigger-tags and distinguish \emph{token-level trigger-tags}, which introduce watermark-inspired signals during decoding, from \emph{weight-level trigger-tags}, which learn backdoor-inspired associations between target conditions and detectable model behavior. Furthermore, (ii)~we introduce \Untag, a unified attack framework that organizes their mechanism-specific attack surfaces into a common taxonomy. We evaluate representative token-level and weight-level trigger-tags using phishing as a case study. We find that while trigger-tags may provide useful evidence in controlled settings, our attacks render the existing trigger-tag mechanisms to be entirely ineffective. Consequently, we argue that these mechanisms should not be treated as robust misuse detectors when attackers can transform outputs or modify open weights.

cs.CR↗

The Tokens Remember: When Tokenization Bypasses Knowledge Editing and Unlearning

Open-weight LLMs give downstream users control over the inference stack, but this flexibility can undermine post-release guarantees that sensitive knowledge has been modified or removed. Model editing and machine unlearning are used to modify or remove targeted knowledge without retraining models from scratch. However, existing security evaluations of these techniques face two critical limitations. First, they typically require access to either the original pre-edit/unlearning model or auxiliary classifiers to detect modifications or reconstruct pre-edit behavior. Second, they evaluate modifications under the canonical tokenization of an input, implicitly treating tokenization as a benign preprocessing step. We show that this assumption creates a security gap: the same input string can be represented by alternative valid tokenizations that induce different computational trajectories, allowing an adversary to bypass localized modifications and recover information intended to be suppressed. We introduce Toketive, a simple yet powerful reference-free attack that exploits the tokenization-based side channel to (i) detect modified knowledge and (ii) reconstruct the corresponding pre-edit response. It operates solely on the released model and requires neither the pre-edit model, training data, shadow models, nor auxiliary classifiers. Across five LLMs, six datasets, and six editing and unlearning techniques, we find that 38.6% of alternative tokenizations bypass the modification and recover the pre-edit response. Toketive detects modified facts with an F1 score of 84.2%, a 26.2% relative gain over the strongest baseline, and reconstructs pre-edit responses with 74.5% top-5 accuracy, 21.7% higher than the best baseline. Our results show that localized modifications should not be treated as robust knowledge-control boundaries without adversarial evaluation over alternative representations.

cs.CR↗

Relevance as a Vulnerability: How Web Retrieval Degrades Safety Alignment in LLM Agents

AI agents augment large language models with external tools such as web retrieval, enabling grounded and up-to-date responses. However, incorporating external content into the generation pipeline can weaken the safety alignment mechanisms that govern model outputs. Prior work shows that enabling retrieval in agents increases compliance with harmful requests. We introduce AgentREVEAL, a diagnostic framework for analyzing retrieval-induced safety degradation in LLM agents. The framework examines two axes: how retrieval is integrated into the agent pipeline and the properties of the retrieved content. Along the integration axis, we find that binding tool invocation and response generation in a single step amplifies harmful outputs. Along the content axis, we uncover the Safe Source Paradox: even oppositional or safety-oriented sources, such as pages containing warnings or risk disclaimers, can increase harmful compliance by an average of 25% compared to the no-retrieval baseline. Finally, we show that relevance acts as a shared activation condition for both vulnerabilities. Similar patterns appear on frontier closed models, and harmful compliance remains elevated under several representative pipeline interventions, with some agents also entering this regime under autonomous retrieval. Because relevance is also what makes retrieval useful, these results expose a safety-utility trade-off for retrieval-enabled agents. We introduce HarmURLBench, a benchmark containing 1,405 real-world URLs paired with 320 harmful behaviors to support future evaluations.

cs.CL↗

ThinkEval: Practical Evaluation of Knowledge Leakage in LLM Editing using Thought-based Knowledge Graphs

Robust model-editing techniques are essential for deploying large language models (LLMs) in practical applications, as they enable cost-effective ways to deal with challenges such as privacy breaches, bias mitigation and misinformation spread. For example, an LLM-based healthcare assistance may need to update out-dated or incorrect knowledge to prevent harmful recommendations. However, many editing techniques focus on isolated facts, which critically fail to prevent indirect knowledge leakage -- the unintended reconstruction of edited-out information through persistent causal links and contextual relationships. To assist users in selecting the right editing technique, we develop and present ThinkEval, a framework to systematically quantify indirect knowledge leakage and ripple effects in model-editing. ThinkEval builds and employs specialized knowledge graphs to analyze the causal structure of facts before and after editing. To support this approach, we present KnowGIC, a benchmark dataset comprising multi-step reasoning paths that precisely measure these complex knowledge transformation effects. We evaluate five editing techniques: AlphaEdit, RECT, ROME, MEMIT, and PRUNE across multiple LLMs. Our results show that these techniques struggle to balance indirect fact suppression with the preservation of related knowledge, compromising the contextual integrity of a model's knowledge. Our dataset is available at: https://github.com/manitbaser/KnowGIC.

cs.LG↗

Context-Aware Information Transfer via Digital Semantic Communication in UAV-Based Networks

In smart cities, bandwidth-constrained Unmanned Aerial Vehicles (UAVs) often fail to relay mission-critical data in time, compromising real-time decision-making. This highlights the need for faster and more efficient transmission of only the most relevant information. To address this, we propose DSC-UAV model, leveraging a context-adaptive Digital Semantic Communication (DSC) framework. This model redefines aerial data transmission through three core components: prompt-aware encoding, dynamic UAV-enabled relaying, and user mobility-optimized reinforcement learning. Ground users transmit context-driven visual content. Images are encoded via Vision Transformer combined with a prompt-text encoder to generate semantic features based on the desired context (generic or object-specific). These features are then quantized and transmitted over a UAV network that dynamically relays the data. Joint trajectory and resource allocation are optimized using Truncated Quantile Critic (TQC)-aided reinforcement learning technique, which offers greater stability and precision over standard SAC and TD3 due to its resistance to overestimation bias. Simulations demonstrate significant performance improvement, up to 22% gain in semantic-structural similarity and 14% reduction in Age of Information (AoI) compared to digital and prior UAV-semantic communication baselines. By integrating mobility control with context-driven visual abstraction, DSC-UAV advances resilient, information-centric surveillance for next-generation UAV networks in bandwidth-constrained environments.

eess.SY↗

RECTor: Robust and Efficient Correlation Attack on Tor

Tor is a widely used anonymity network that conceals user identities by routing traffic through encrypted relays, yet it remains vulnerable to traffic correlation attacks that deanonymize users by matching patterns in ingress and egress traffic. However, existing correlation methods suffer from two major limitations: limited robustness to noise and partial observations, and poor scalability due to computationally expensive pairwise matching. To address these challenges, we propose RECTor, a machine learning-based framework for traffic correlation under realistic conditions. RECTor employs attention-based Multiple Instance Learning (MIL) and GRU-based temporal encoding to extract robust flow representations, even when traffic data is incomplete or obfuscated. These embeddings are mapped into a shared space via a Siamese network and efficiently matched using approximate nearest neighbor (aNN) search. Empirical evaluations show that RECTor outperforms state-of-the-art baselines such as DeepCorr, DeepCOFFEA, and FlowTracker, achieving up to 60% higher true positive rates under high-noise conditions and reducing training and inference time by over 50%. Moreover, RECTor demonstrates strong scalability: inference cost grows near-linearly as the number of flows increases. These findings reveal critical vulnerabilities in Tor's anonymity model and highlight the need for advanced model-aware defenses.

cs.CR↗

Time Series Based Network Intrusion Detection using MTF-Aided Transformer

This paper introduces a novel approach to time series classification using a Markov Transition Field (MTF)-aided Transformer model, specifically designed for Software-Defined Networks (SDNs). The proposed model integrates the temporal dependency modeling strengths of MTFs with the sophisticated pattern recognition capabilities of Transformer architectures. We evaluate the model's performance using the InSDN dataset, demonstrating that our model outperforms baseline classification models, particularly in data-constrained environments commonly encountered in SDN applications. We also highlight the relationship between the MTF and Transformer components, which leads to better performance, even with limited data. Furthermore, our approach achieves competitive training and inference times, making it an efficient solution for real-world SDN applications. These findings establish the potential of MTF-aided Transformers to address the challenges of time series classification in SDNs, offering a promising path for reliable and scalable analysis in scenarios with sparse data.

cs.NI↗

UniNet: A Unified Multi-granular Traffic Modeling Framework for Network Security

As modern networks grow increasingly complex--driven by diverse devices, encrypted protocols, and evolving threats--network traffic analysis has become critically important. Existing machine learning models often rely only on a single representation of packets or flows, limiting their ability to capture the contextual relationships essential for robust analysis. Furthermore, task-specific architectures for supervised, semi-supervised, and unsupervised learning lead to inefficiencies in adapting to varying data formats and security tasks. To address these gaps, we propose UniNet, a unified framework that introduces a novel multi-granular traffic representation (T-Matrix), integrating session, flow, and packet-level features to provide comprehensive contextual information. Combined with T-Attent, a lightweight attention-based model, UniNet efficiently learns latent embeddings for diverse security tasks. Extensive evaluations across four key network security and privacy problems--anomaly detection, attack classification, IoT device identification, and encrypted website fingerprinting--demonstrate UniNet's significant performance gain over state-of-the-art methods, achieving higher accuracy, lower false positive rates, and improved scalability. By addressing the limitations of single-level models and unifying traffic analysis paradigms, UniNet sets a new benchmark for modern network security.

cs.CR↗

Securing the Skies: An IRS-Assisted AoI-Aware Secure Multi-UAV System with Efficient Task Offloading

Unmanned Aerial Vehicles (UAVs) are integral in various sectors like agriculture, surveillance, and logistics, driven by advancements in 5G. However, existing research lacks a comprehensive approach addressing both data freshness and security concerns. In this paper, we address the intricate challenges of data freshness, and security, especially in the context of eavesdropping and jamming in modern UAV networks. Our framework incorporates exponential AoI metrics and emphasizes secrecy rate to tackle eavesdropping and jamming threats. We introduce a transformer-enhanced Deep Reinforcement Learning (DRL) approach to optimize task offloading processes. Comparative analysis with existing algorithms showcases the superiority of our scheme, indicating its promising advancements in UAV network management.

eess.SY↗

ZEST: Attention-based Zero-Shot Learning for Unseen IoT Device Classification

Recent research works have proposed machine learning models for classifying IoT devices connected to a network. However, there is still a practical challenge of not having all devices (and hence their traffic) available during the training of a model. This essentially means, during the operational phase, we need to classify new devices not seen in the training phase. To address this challenge, we propose ZEST -- a ZSL (zero-shot learning) framework based on self-attention for classifying both seen and unseen devices. ZEST consists of i) a self-attention based network feature extractor, termed SANE, for extracting latent space representations of IoT traffic, ii) a generative model that trains a decoder using latent features to generate pseudo data, and iii) a supervised model that is trained on the generated pseudo data for classifying devices. We carry out extensive experiments on real IoT traffic data; our experiments demonstrate i) ZEST achieves significant improvement (in terms of accuracy) over the baselines; ii) SANE is able to better extract meaningful representations than LSTM which has been commonly used for modeling network traffic.

cs.NI↗

Reliable and Efficient Data Collection in UAV-based IoT Networks

Internet of Things (IoT) involves sensors for monitoring and wireless networks for efficient communication. However, resource-constrained IoT devices and limitations in existing wireless technologies hinder its full potential. Integrating Unmanned Aerial Vehicles (UAVs) into IoT networks can address some challenges by expanding its' coverage, providing security, and bringing computing closer to IoT devices. Nevertheless, effective data collection in UAV-assisted IoT networks is hampered by factors, including dynamic UAV behavior, environmental variables, connectivity instability, and security considerations. In this survey, we first explore UAV-based IoT networks, focusing on communication and networking aspects. Next, we cover various UAV-based data collection methods their advantages and disadvantages, followed by a discussion on performance metrics for data collection. As this article primarily emphasizes reliable and efficient data collection in UAV-assisted IoT networks, we briefly discuss existing research on data accuracy and consistency, network connectivity, and data security and privacy to provide insights into reliable data collection. Additionally, we discuss efficient data collection strategies in UAV-based IoT networks, covering trajectory and path planning, collision avoidance, sensor network clustering, data aggregation, UAV swarm formations, and artificial intelligence for optimization. We also present two use cases of UAVs as a service for enhancing data collection reliability and efficiency. Finally, we discuss future challenges in data collection for UAV-assisted IoT networks.

cs.NI↗

DOSM: Demand-Prediction based Online Service Management for Vehicular Edge Computing Networks

In this work, we investigate an online service management problem in vehicular edge computing networks. To satisfy the varying service demands of mobile vehicles, a service management framework is required to make decisions on the service lifecycle to maintain good network performance. We describe the service lifecycle consists of creating an instance of a given service (\textit{scale-out}), moving an instance to a different edge node (\textit{migration}), and/or termination of an underutilized instance (\textit{scale-in}). In this paper, we propose an efficient online algorithm to perform service management in each time slot, where performance quality in the current time slot, the service demand in future time slots, and the minimal observed delay by vehicles and the minimal migration delay are considered while making the decisions on service lifecycle. Here, the future service demand is computed from a gated recurrent unit (GRU)-based prediction model, and the network performance quality is estimated using a deep reinforcement learning (DRL) model which has the ability to interact with the vehicular environment in real-time. The choice of optimal edge location to deploy a service instance at different times is based on our proposed optimization formulations. Simulation experiments using real-world vehicle trajectories are carried out to evaluate the performance of our proposed demand-prediction based online service management (DOSM) framework against different state-of-the-art solutions using several performance metrics.

cs.NI↗

Optimizing Vehicle-to-Edge Mapping with Load Balancing for Attack-Resilience in IoV

Attack-resilience is essential to maintain continuous service availability in Internet of Vehicles (IoV) where critical tasks are carried out. In this paper, we address the problem of service outage due to attacks on the edge network and propose an attack-resilient mapping of vehicles to edge nodes that host different types of service instances considering resource efficiency and delay. The distribution of service requests (of an attack-affected edge node) to multiple attack-free edge nodes is performed with an optimal vehicle-to-edge (V2E) mapping. The optimal mapping aims to improve the user experience with minimal delay while considering fair usage of edge capacities and balanced load upon a failure over different edge nodes. The proposed mapping solution is used within a deep reinforcement learning (DRL) based framework to effectively deal with the dynamism in service requests and vehicle mobility. We demonstrate the effectiveness of the proposed mapping approach through extensive simulation results using real-world vehicle mobility datasets from three cities.

cs.NI↗