arXiv subjects

Alexandra Dmitrienko

Publications and source records attributed to Alexandra Dmitrienko.

At least 19 recordsLinked to original sources

arXiv2026

CodePoisonRAG: Knowledge Poisoning Attacks on Retrieval-Augmented Code Generation

Retrieval-Augmented Code Generation (RACG) improves LLM-based software development by retrieving external code artifacts, documentation, and patches, and incorporating them into the generation context. This reliance on external knowledge introduces a critical trust boundary: poisoned artifacts can influence generated code without modifying the underlying LLM. Prior work shows that selecting existing vulnerable examples can increase the general vulnerability rate of RACG outputs, but leaves open whether a black-box attacker can construct a single task-matched artifact that propagates an attacker-selected weakness. We introduce CodePoisonRAG, a targeted upstream knowledge-poisoning framework that transforms benign fixed-code entries into poisoned artifacts. Its attack chain combines CWE-specific Vulnerability Injection, which embeds a selected source-to-sink flow while retaining task alignment, with Semantic Mislabeling, which adds false safety claims without repairing the vulnerable behavior. The attacker has no access to the victim's deployed knowledge base, retriever, re-ranker, generator, prompt, or defense mechanism and injects at most one artifact per anticipated programming task. We construct 85 poisoned artifacts covering ten CWE classes across Java and C, yielding an aggregate corpus-poisoning ratio of 0.7%. Across three generators, all 85 artifacts appear among the Top-3 results for their corresponding queries, and CodePoisonRAG achieves attack success rates between 0.80 and 0.93. Against CodeGuarder, which injects vulnerability-specific security knowledge into the generation context, the attack retains success rates between 0.40 and 0.71. These results show that RACG poisoning extends beyond the incidental propagation of existing vulnerabilities to the targeted construction and propagation of attacker-selected weaknesses.

cs.CR↗

arXiv2026

RAVEN: Agentic RAG for Automated Vulnerability Repair

Automated vulnerability repair has emerged as a promising direction to mitigate the growing number of software vulnerabilities. Recent advances in Large Language Models (LLMs) have further accelerated research in automated repair. However, existing frameworks remain largely restricted to memory-related vulnerabilities and locally repairable vulnerability settings, leaving generalization to unseen vulnerability types underexplored. Their evaluations are often limited to a single programming language, and largely rely on proprietary models. In this paper, we propose RAVEN, a scalable, efficient and autonomous framework that integrates an agentic retrieval-augmented generation (RAG) pipeline with controlled iterative repair in a unified framework. The framework utilizes open-source LLMs in a fully locally deployable setting with limited GPU requirements, while building a multi-faceted retrieval pipeline to retrieve historically relevant vulnerability fixes and guide the patch generation. In addition, RAVEN introduces a dedicated Curator Agent that retrieves cross-file dependencies from the target repository, to fix complex vulnerabilities that cannot be addressed using local vulnerable code alone. We evaluate RAVEN on 160 real-world CVE vulnerabilities across diverse vulnerability types, two programming languages, unseen CWE categories, and out-of-distribution settings. RAVEN achieves an overall repair success rate of 83.13%, outperforming all existing state-of-the-art repair frameworks, while also demonstrating strong generalization capabilities and maintaining the repair cost negligible.

cs.CR↗

arXiv2025

TwinBreak: Jailbreaking LLM Security Alignments based on Twin Prompts

Machine learning is advancing rapidly, with applications bringing notable benefits, such as improvements in translation and code generation. Models like ChatGPT, powered by Large Language Models (LLMs), are increasingly integrated into daily life. However, alongside these benefits, LLMs also introduce social risks. Malicious users can exploit LLMs by submitting harmful prompts, such as requesting instructions for illegal activities. To mitigate this, models often include a security mechanism that automatically rejects such harmful prompts. However, they can be bypassed through LLM jailbreaks. Current jailbreaks often require significant manual effort, high computational costs, or result in excessive model modifications that may degrade regular utility. We introduce TwinBreak, an innovative safety alignment removal method. Building on the idea that the safety mechanism operates like an embedded backdoor, TwinBreak identifies and prunes parameters responsible for this functionality. By focusing on the most relevant model layers, TwinBreak performs fine-grained analysis of parameters essential to model utility and safety. TwinBreak is the first method to analyze intermediate outputs from prompts with high structural and content similarity to isolate safety parameters. We present the TwinPrompt dataset containing 100 such twin prompts. Experiments confirm TwinBreak's effectiveness, achieving 89% to 98% success rates with minimal computational requirements across 16 LLMs from five vendors.

cs.LG↗

arXiv2024

Memory Backdoor Attacks on Neural Networks

Neural networks are often trained on proprietary datasets, making them attractive attack targets. We present a novel dataset extraction method leveraging an innovative training time backdoor attack, allowing a malicious federated learning server to systematically and deterministically extract complete client training samples through a simple indexing process. Unlike prior techniques, our approach guarantees exact data recovery rather than probabilistic reconstructions or hallucinations, provides precise control over which samples are memorized and how many, and shows high capacity and robustness. Infected models output data samples when they receive a patternbased index trigger, enabling systematic extraction of meaningful patches from each clients local data without disrupting global model utility. To address small model output sizes, we extract patches and then recombined them. The attack requires only a minor modification to the training code that can easily evade detection during client-side verification. Hence, this vulnerability represents a realistic FL supply-chain threat, where a malicious server can distribute modified training code to clients and later recover private data from their updates. Evaluations across classifiers, segmentation models, and large language models demonstrate that thousands of sensitive training samples can be recovered from client models with minimal impact on task performance, and a clients entire dataset can be stolen after multiple FL rounds. For instance, a medical segmentation dataset can be extracted with only a 3 percent utility drop. These findings expose a critical privacy vulnerability in FL systems, emphasizing the need for stronger integrity and transparency in distributed training pipelines.

cs.CR↗

arXiv2024

GNN-Based Code Annotation Logic for Establishing Security Boundaries in C Code

Securing sensitive operations in today's interconnected software landscape is crucial yet challenging. Modern platforms rely on Trusted Execution Environments (TEEs), such as Intel SGX and ARM TrustZone, to isolate security sensitive code from the main system, reducing the Trusted Computing Base (TCB) and providing stronger assurances. However, identifying which code should reside in TEEs is complex and requires specialized expertise, which is not supported by current automated tools. Existing solutions often migrate entire applications to TEEs, leading to suboptimal use and an increased TCB. To address this gap, we propose Code Annotation Logic (CAL), a pioneering tool that automatically identifies security sensitive components for TEE isolation. CAL analyzes codebases, leveraging a graph-based approach with novel feature construction and employing a custom graph neural network model to accurately determine which parts of the code should be isolated. CAL effectively optimizes TCB, reducing the burden of manual analysis and enhancing overall security. Our contributions include the definition of security sensitive code, the construction and labeling of a comprehensive dataset of source files, a feature rich graph based data preparation pipeline, and the CAL model for TEE integration. Evaluation results demonstrate CAL's efficacy in identifying sensitive code with a recall of 86.05%, an F1 score of 81.56%, and an identification rate of 91.59% for security sensitive functions. By enabling efficient code isolation, CAL advances the secure development of applications using TEEs, offering a practical solution for developers to reduce attack vectors.

cs.CR↗

arXiv2024

SPOQchain: Platform for Secure, Scalable, and Privacy-Preserving Supply Chain Tracing and Counterfeit Protection

Product lifecycle tracing is increasingly in the focus of regulators and producers, as shown with the initiative of the Digital Product Pass. Likewise, new methods of counterfeit detection are developed that are, e.g., based on Physical Unclonable Functions (PUFs). In order to ensure trust and integrity of product lifecycle data, multiple existing supply chain tracing systems are built on blockchain technology. However, only few solutions employ secure identifiers such as PUFs. Furthermore, existing systems that publish the data of individual products, in part fully transparently, have a detrimental impact on scalability and the privacy of users. This work proposes SPOQchain, a novel blockchain-based platform that provides comprehensive lifecycle traceability and originality verification while ensuring high efficiency and user privacy. The improved efficiency is achieved by a sophisticated batching mechanism that removes lifecycle redundancies. In addition to the successful evaluation of SPOQchain's scalability, this work provides a comprehensive analysis of privacy and security aspects, demonstrating the need and qualification of SPOQchain for the future of supply chain tracing.

cs.CR↗

arXiv2024

Time-Aware Face Anti-Spoofing with Rotation Invariant Local Binary Patterns and Deep Learning

Facial recognition systems have become an integral part of the modern world. These methods accomplish the task of human identification in an automatic, fast, and non-interfering way. Past research has uncovered high vulnerability to simple imitation attacks that could lead to erroneous identification and subsequent authentication of attackers. Similar to face recognition, imitation attacks can also be detected with Machine Learning. Attack detection systems use a variety of facial features and advanced machine learning models for uncovering the presence of attacks. In this work, we assess existing work on liveness detection and propose a novel approach that promises high classification accuracy by combining previously unused features with time-aware deep learning strategies.

cs.CV↗

arXiv2024

DNNShield: Embedding Identifiers for Deep Neural Network Ownership Verification

The surge in popularity of machine learning (ML) has driven significant investments in training Deep Neural Networks (DNNs). However, these models that require resource-intensive training are vulnerable to theft and unauthorized use. This paper addresses this challenge by introducing DNNShield, a novel approach for DNN protection that integrates seamlessly before training. DNNShield embeds unique identifiers within the model architecture using specialized protection layers. These layers enable secure training and deployment while offering high resilience against various attacks, including fine-tuning, pruning, and adaptive adversarial attacks. Notably, our approach achieves this security with minimal performance and computational overhead (less than 5\% runtime increase). We validate the effectiveness and efficiency of DNNShield through extensive evaluations across three datasets and four model architectures. This practical solution empowers developers to protect their DNNs and intellectual property rights.

cs.CR↗

arXiv2023

Vulnerability Scanners for Ethereum Smart Contracts: A Large-Scale Study

Ethereum smart contracts, which are autonomous decentralized applications on the blockchain that manage assets often exceeding millions of dollars, have become primary targets for cyberattacks. In 2023 alone, such vulnerabilities led to substantial financial losses exceeding a billion of US dollars. To counter these threats, various tools have been developed by academic and commercial entities to detect and mitigate vulnerabilities in smart contracts. Our study investigates the gap between the effectiveness of existing security scanners and the vulnerabilities that still persist in practice. We compiled four distinct datasets for this analysis. The first dataset comprises 77,219 source codes extracted directly from the blockchain, while the second includes over 4 million bytecodes obtained from Ethereum Mainnet and testnets. The other two datasets consist of nearly 14,000 manually annotated smart contracts and 373 smart contracts verified through audits, providing a foundation for a rigorous ground truth analysis on bytecode and source code. Using the unlabeled datasets, we conducted a comprehensive quantitative evaluation of 17 vulnerability scanners, revealing considerable discrepancies in their findings. Our analysis of the ground truth datasets indicated poor performance across all the tools we tested. This study unveils the reasons for poor performance and underscores that the current state of the art for smart contract security falls short in effectively addressing open problems, highlighting that the challenge of effectively detecting vulnerabilities remains a significant and unresolved issue.

cs.CR↗

arXiv2023

FreqFed: A Frequency Analysis-Based Approach for Mitigating Poisoning Attacks in Federated Learning

Federated learning (FL) is a collaborative learning paradigm allowing multiple clients to jointly train a model without sharing their training data. However, FL is susceptible to poisoning attacks, in which the adversary injects manipulated model updates into the federated model aggregation process to corrupt or destroy predictions (untargeted poisoning) or implant hidden functionalities (targeted poisoning or backdoors). Existing defenses against poisoning attacks in FL have several limitations, such as relying on specific assumptions about attack types and strategies or data distributions or not sufficiently robust against advanced injection techniques and strategies and simultaneously maintaining the utility of the aggregated model. To address the deficiencies of existing defenses, we take a generic and completely different approach to detect poisoning (targeted and untargeted) attacks. We present FreqFed, a novel aggregation mechanism that transforms the model updates (i.e., weights) into the frequency domain, where we can identify the core frequency components that inherit sufficient information about weights. This allows us to effectively filter out malicious updates during local training on the clients, regardless of attack types, strategies, and clients' data distributions. We extensively evaluate the efficiency and effectiveness of FreqFed in different application domains, including image classification, word prediction, IoT intrusion detection, and speech recognition. We demonstrate that FreqFed can mitigate poisoning attacks effectively with a negligible impact on the utility of the aggregated model.

cs.CR↗

arXiv2023

ClearMark: Intuitive and Robust Model Watermarking via Transposed Model Training

Due to costly efforts during data acquisition and model training, Deep Neural Networks (DNNs) belong to the intellectual property of the model creator. Hence, unauthorized use, theft, or modification may lead to legal repercussions. Existing DNN watermarking methods for ownership proof are often non-intuitive, embed human-invisible marks, require trust in algorithmic assessment that lacks human-understandable attributes, and rely on rigid thresholds, making it susceptible to failure in cases of partial watermark erasure. This paper introduces ClearMark, the first DNN watermarking method designed for intuitive human assessment. ClearMark embeds visible watermarks, enabling human decision-making without rigid value thresholds while allowing technology-assisted evaluations. ClearMark defines a transposed model architecture allowing to use of the model in a backward fashion to interwove the watermark with the main task within all model parameters. Compared to existing watermarking methods, ClearMark produces visual watermarks that are easy for humans to understand without requiring complex verification algorithms or strict thresholds. The watermark is embedded within all model parameters and entangled with the main task, exhibiting superior robustness. It shows an 8,544-bit watermark capacity comparable to the strongest existing work. Crucially, ClearMark's effectiveness is model and dataset-agnostic, and resilient against adversarial model manipulations, as demonstrated in a comprehensive study performed with four datasets and seven architectures.

cs.LG↗

arXiv2023

Metadata-based Malware Detection on Android using Machine Learning

In the digitized world, smartphones and their apps play an important role. To name just a few examples, some apps offer possibilities for entertainment, others for online banking, and others offer support for two-factor authentication. Therefore, with smartphones also, sensitive information is shared; thus, they are a desirable target for malware. The following technical report gives an overview of how machine learning, especially neural networks, can be employed to detect malicious Android apps based on their metadata. Detection based on the metadata is necessary since not all of an app's information is readable from another app due to the security layout of Android. To do so, a comparable big dataset of metadata of apps has been collected for learning and evaluation in this work. The first section, after the introduction, presents the related work, followed by the description of the sources of the dataset and the selection of the features used for machine learning, in this case, only the app permissions. Afterward, a free available dataset is used to find an efficient and effective neural network model for learning and evaluation. Here, the fully connected network type consisting of dense layers is chosen. Then this model is trained and evaluated on the new, more extensive dataset to obtain a representative result. It turns out that this model detects malware with an accuracy of 92.93% based on an app's permissions.

cs.CR↗

arXiv2023

G-Scan: Graph Neural Networks for Line-Level Vulnerability Identification in Smart Contracts

Due to the immutable and decentralized nature of Ethereum (ETH) platform, smart contracts are prone to security risks that can result in financial loss. While existing machine learning-based vulnerability detection algorithms achieve high accuracy at the contract level, they require developers to manually inspect source code to locate bugs. To this end, we present G-Scan, the first end-to-end fine-grained line-level vulnerability detection system evaluated on the first-of-its-kind real world dataset. G-Scan first converts smart contracts to code graphs in a dependency and hierarchy preserving manner. Next, we train a graph neural network to identify vulnerable nodes and assess security risks. Finally, the code graphs with node vulnerability predictions are mapped back to the smart contracts for line-level localization. We train and evaluate G-Scan on a collected real world smart contracts dataset with line-level annotations on reentrancy vulnerability, one of the most common and severe types of smart contract vulnerabilities. With the well-designed graph representation and high-quality dataset, G-Scan achieves 93.02% F1-score in contract-level vulnerability detection and 93.69% F1-score in line-level vulnerability localization. Additionally, the lightweight graph neural network enables G-Scan to localize vulnerabilities in 6.1k lines of code smart contract within 1.2 seconds.

cs.CR↗

arXiv2023

TorMult: Introducing a Novel Tor Bandwidth Inflation Attack

The Tor network is the most prominent system for providing anonymous communication to web users, with a daily user base of 2 million users. However, since its inception, it has been constantly targeted by various traffic fingerprinting and correlation attacks aiming at deanonymizing its users. A critical requirement for these attacks is to attract as much user traffic to adversarial relays as possible, which is typically accomplished by means of bandwidth inflation attacks. This paper proposes a new inflation attack vector in Tor, referred to as TorMult, which enables inflation of measured bandwidth. The underlying attack technique exploits resource sharing among Tor relay nodes and employs a cluster of attacker-controlled relays with coordinated resource allocation within the cluster to deceive bandwidth measurers into believing that each relay node in the cluster possesses ample resources. We propose two attack variants, C-TorMult and D-TorMult, and test both versions in a private Tor test network. Our evaluation demonstrates that an attacker can inflate the measured bandwidth by a factor close to n using C-TorMult and nearly half n*N using D-TorMult, where n is the size of the cluster hosted on one server and N is the number of servers. Furthermore, our theoretical analysis reveals that gaining control over half of the Tor network's traffic can be achieved by employing just 10 dedicated servers with a cluster size of 109 relays running the TorMult attack, each with a bandwidth of 100MB/s. The problem is further exacerbated by the fact that Tor not only allows resource sharing but, according to recent reports, even promotes it.

cs.CR↗

arXiv2023

Avoid Adversarial Adaption in Federated Learning by Multi-Metric Investigations

Federated Learning (FL) facilitates decentralized machine learning model training, preserving data privacy, lowering communication costs, and boosting model performance through diversified data sources. Yet, FL faces vulnerabilities such as poisoning attacks, undermining model integrity with both untargeted performance degradation and targeted backdoor attacks. Preventing backdoors proves especially challenging due to their stealthy nature. Prominent mitigation techniques against poisoning attacks rely on monitoring certain metrics and filtering malicious model updates. While shown effective in evaluations, we argue that previous works didn't consider realistic real-world adversaries and data distributions. We define a new notion of strong adaptive adversaries, capable of adapting to multiple objectives simultaneously. Through extensive empirical tests, we show that existing defense methods can be easily circumvented in this adversary model. We also demonstrate, that existing defenses have limited effectiveness when no assumptions are made about underlying data distributions. We introduce Metric-Cascades (MESAS), a novel defense method for more realistic scenarios and adversary models. MESAS employs multiple detection metrics simultaneously to identify poisoned model updates, creating a complex multi-objective optimization problem for adaptive attackers. In our extensive evaluation featuring nine backdoors and three datasets, MESAS consistently detects even strong adaptive attackers. Furthermore, MESAS outperforms existing defenses in distinguishing backdoors from data distribution-related distortions within and across clients. MESAS is the first defense robust against strong adaptive adversaries, effective in real-world data scenarios, with an average overhead of just 24.37 seconds.

cs.LG↗

arXiv2023

AuthentiSense: A Scalable Behavioral Biometrics Authentication Scheme using Few-Shot Learning for Mobile Platforms

Mobile applications are widely used for online services sharing a large amount of personal data online. One-time authentication techniques such as passwords and physiological biometrics (e.g., fingerprint, face, and iris) have their own advantages but also disadvantages since they can be stolen or emulated, and do not prevent access to the underlying device, once it is unlocked. To address these challenges, complementary authentication systems based on behavioural biometrics have emerged. The goal is to continuously profile users based on their interaction with the mobile device. However, existing behavioural authentication schemes are not (i) user-agnostic meaning that they cannot dynamically handle changes in the user-base without model re-training, or (ii) do not scale well to authenticate millions of users. In this paper, we present AuthentiSense, a user-agnostic, scalable, and efficient behavioural biometrics authentication system that enables continuous authentication and utilizes only motion patterns (i.e., accelerometer, gyroscope and magnetometer data) while users interact with mobile apps. Our approach requires neither manually engineered features nor a significant amount of data for model training. We leverage a few-shot learning technique, called Siamese network, to authenticate users at a large scale. We perform a systematic measurement study and report the impact of the parameters such as interaction time needed for authentication and n-shot verification (comparison with enrollment samples) at the recognition stage. Remarkably, AuthentiSense achieves high accuracy of up to 97% in terms of F1-score even when evaluated in a few-shot fashion that requires only a few behaviour samples per user (3 shots). Our approach accurately authenticates users only after 1 second of user interaction. For AuthentiSense, we report a FAR and FRR of 0.023 and 0.057, respectively.

cs.CR↗

arXiv2022

CrowdGuard: Federated Backdoor Detection in Federated Learning

Federated Learning (FL) is a promising approach enabling multiple clients to train Deep Neural Networks (DNNs) collaboratively without sharing their local training data. However, FL is susceptible to backdoor (or targeted poisoning) attacks. These attacks are initiated by malicious clients who seek to compromise the learning process by introducing specific behaviors into the learned model that can be triggered by carefully crafted inputs. Existing FL safeguards have various limitations: They are restricted to specific data distributions or reduce the global model accuracy due to excluding benign models or adding noise, are vulnerable to adaptive defense-aware adversaries, or require the server to access local models, allowing data inference attacks. This paper presents a novel defense mechanism, CrowdGuard, that effectively mitigates backdoor attacks in FL and overcomes the deficiencies of existing techniques. It leverages clients' feedback on individual models, analyzes the behavior of neurons in hidden layers, and eliminates poisoned models through an iterative pruning scheme. CrowdGuard employs a server-located stacked clustering scheme to enhance its resilience to rogue client feedback. The evaluation results demonstrate that CrowdGuard achieves a 100% True-Positive-Rate and True-Negative-Rate across various scenarios, including IID and non-IID data distributions. Additionally, CrowdGuard withstands adaptive adversaries while preserving the original performance of protected models. To ensure confidentiality, CrowdGuard uses a secure and privacy-preserving architecture leveraging Trusted Execution Environments (TEEs) on both client and server sides.

cs.CR↗

arXiv2022

Digital Contact Tracing Solutions: Promises, Pitfalls and Challenges

The COVID-19 pandemic has caused many countries to deploy novel digital contact tracing (DCT) systems to boost the efficiency of manual tracing of infection chains. In this paper, we systematically analyze DCT solutions and categorize them based on their design approaches and architectures. We analyze them with regard to effectiveness, security, privacy, and ethical aspects and compare prominent solutions with regard to these requirements. In particular, we discuss the shortcomings of the Google and Apple Exposure Notification API (GAEN) that is currently widely adopted all over the world. We find that the security and privacy of GAEN have considerable deficiencies as it can be compromised by severe, large-scale attacks. We also discuss other proposed approaches for contact tracing, including our proposal TRACECORONA, that are based on Diffie-Hellman (DH) key exchange and aim at tackling shortcomings of existing solutions. Our extensive analysis shows thatTRACECORONA fulfills the above security requirements better than deployed state-of-the-art approaches. We have implementedTRACECORONA, and its beta test version has been used by more than 2000 users without any major functional problems, demonstrating that there are no technical reasons requiring to make compromises with regard to the requirements of DCTapproaches.

cs.CR↗