arXiv subjects

Stjepan Picek

Publications and source records attributed to Stjepan Picek.

At least 55 records · Page 3Linked to original sources

arXiv2025

Hardware-Enabled Mechanisms for Verifying Responsible AI Development

Advancements in AI capabilities, driven in large part by scaling up computing resources used for AI training, have created opportunities to address major global challenges but also pose risks of misuse. Hardware-enabled mechanisms (HEMs) can support responsible AI development by enabling verifiable reporting of key properties of AI training activities such as quantity of compute used, training cluster configuration or location, as well as policy enforcement. Such tools can promote transparency and improve security, while addressing privacy and intellectual property concerns. Based on insights from an interdisciplinary workshop, we identify open questions regarding potential implementation approaches, emphasizing the need for further research to ensure robust, scalable solutions.

cs.CR↗

arXiv2025

Degree is Important: On Evolving Homogeneous Boolean Functions

Boolean functions with good cryptographic properties like high nonlinearity and algebraic degree play an important in the security of stream and block ciphers. Such functions may be designed, for instance, by algebraic constructions or metaheuristics. This paper investigates the use of Evolutionary Algorithms (EAs) to design homogeneous bent Boolean functions, i.e., functions that are maximally nonlinear and whose algebraic normal form contains only monomials of the same degree. In our work, we evaluate three genotype encodings and four fitness functions. Our results show that while EAs manage to find quadratic homogeneous bent functions (with the best method being a GA leveraging a restricted encoding), none of the approaches result in cubic homogeneous bent functions.

cs.NE↗

arXiv2024

More is Better (Mostly): On the Backdoor Attacks in Federated Graph Neural Networks

Graph Neural Networks (GNNs) are a class of deep learning-based methods for processing graph domain information. GNNs have recently become a widely used graph analysis method due to their superior ability to learn representations for complex graph data. However, due to privacy concerns and regulation restrictions, centralized GNNs can be difficult to apply to data-sensitive scenarios. Federated learning (FL) is an emerging technology developed for privacy-preserving settings when several parties need to train a shared global model collaboratively. Although several research works have applied FL to train GNNs (Federated GNNs), there is no research on their robustness to backdoor attacks. This paper bridges this gap by conducting two types of backdoor attacks in Federated GNNs: centralized backdoor attacks (CBA) and distributed backdoor attacks (DBA). Our experiments show that the DBA attack success rate is higher than CBA in almost all evaluated cases. For CBA, the attack success rate of all local triggers is similar to the global trigger even if the training set of the adversarial party is embedded with the global trigger. To further explore the properties of two backdoor attacks in Federated GNNs, we evaluate the attack performance for a different number of clients, trigger sizes, poisoning intensities, and trigger densities. Moreover, we explore the robustness of DBA and CBA against one defense. We find that both attacks are robust against the investigated defense, necessitating the need to consider backdoor attacks in Federated GNNs as a novel threat that requires custom defenses.

cs.CR↗

arXiv2024

The More the Merrier: On Evolving Five-valued Spectra Boolean Functions

Evolving Boolean functions with specific properties is an interesting optimization problem since, depending on the combination of properties and Boolean function size, the problem can range from very simple to (almost) impossible to solve. Moreover, some problems are more interesting as there may be only a few options for generating the required Boolean functions. This paper investigates one such problem: evolving five-valued spectra Boolean functions, which are the functions whose Walsh-Hadamard coefficients can only take five distinct values. We experimented with three solution encodings, two fitness functions, and 12 Boolean function sizes and showed that the tree encoding is superior to other choices, as we can obtain five-valued Boolean functions with high nonlinearity.

cs.NE↗

arXiv2024

BAN: Detecting Backdoors Activated by Adversarial Neuron Noise

Backdoor attacks on deep learning represent a recent threat that has gained significant attention in the research community. Backdoor defenses are mainly based on backdoor inversion, which has been shown to be generic, model-agnostic, and applicable to practical threat scenarios. State-of-the-art backdoor inversion recovers a mask in the feature space to locate prominent backdoor features, where benign and backdoor features can be disentangled. However, it suffers from high computational overhead, and we also find that it overly relies on prominent backdoor features that are highly distinguishable from benign features. To tackle these shortcomings, this paper improves backdoor feature inversion for backdoor detection by incorporating extra neuron activation information. In particular, we adversarially increase the loss of backdoored models with respect to weights to activate the backdoor effect, based on which we can easily differentiate backdoored and clean models. Experimental results demonstrate our defense, BAN, is 1.37$\times$ (on CIFAR-10) and 5.11$\times$ (on ImageNet200) more efficient with an average 9.99\% higher detect success rate than the state-of-the-art defense BTI-DBF. Our code and trained models are publicly available at~\url{https://github.com/xiaoyunxxy/ban}.

cs.LG↗

arXiv2024

Flashy Backdoor: Real-world Environment Backdoor Attack on SNNs with DVS Cameras

While security vulnerabilities in traditional Deep Neural Networks (DNNs) have been extensively studied, the susceptibility of Spiking Neural Networks (SNNs) to adversarial attacks remains mostly underexplored. Until now, the mechanisms to inject backdoors into SNN models have been limited to digital scenarios; thus, we present the first evaluation of backdoor attacks in real-world environments. We begin by assessing the applicability of existing digital backdoor attacks and identifying their limitations for deployment in physical environments. To address each of the found limitations, we present three novel backdoor attack methods on SNNs, i.e., Framed, Strobing, and Flashy Backdoor. We also assess the effectiveness of traditional backdoor procedures and defenses adapted for SNNs, such as pruning, fine-tuning, and fine-pruning. The results show that while these procedures and defenses can mitigate some attacks, they often fail against stronger methods like Flashy Backdoor or sacrifice too much clean accuracy, rendering the models unusable. Overall, all our methods can achieve up to a 100% Attack Success Rate while maintaining high clean accuracy in every tested dataset. Additionally, we evaluate the stealthiness of the triggers with commonly used metrics, finding them highly stealthy. Thus, we propose new alternatives more suited for identifying poisoned samples in these scenarios. Our results show that further research is needed to ensure the security of SNN-based systems against backdoor attacks and their safe application in real-world scenarios. The code, experiments, and results are available in our repository.

cs.CR↗

arXiv2024

$$\mathbf{L^2\cdot M = C^2}$$ Large Language Models are Covert Channels

Large Language Models (LLMs) have gained significant popularity recently. LLMs are susceptible to various attacks but can also improve the security of diverse systems. However, besides enabling more secure systems, how well do open source LLMs behave as covertext distributions to, e.g., facilitate censorship-resistant communication? In this paper, we explore open-source LLM-based covert channels. We empirically measure the security vs. capacity of an open-source LLM model (Llama-7B) to assess its performance as a covert channel. Although our results indicate that such channels are not likely to achieve high practical bitrates, we also show that the chance for an adversary to detect covert communication is low. To ensure our results can be used with the least effort as a general reference, we employ a conceptually simple and concise scheme and only assume public models.

cs.CR↗

arXiv2024

Membership Privacy Evaluation in Deep Spiking Neural Networks

Artificial Neural Networks (ANNs), commonly mimicking neurons with non-linear functions to output floating-point numbers, consistently receive the same signals of a data point during its forward time. Unlike ANNs, Spiking Neural Networks (SNNs) get various input signals in the forward time of a data point and simulate neurons in a biologically plausible way, i.e., producing a spike (a binary value) if the accumulated membrane potential of a neuron is larger than a threshold. Even though ANNs have achieved remarkable success in multiple tasks, e.g., face recognition and object detection, SNNs have recently obtained attention due to their low power consumption, fast inference, and event-driven properties. While privacy threats against ANNs are widely explored, much less work has been done on SNNs. For instance, it is well-known that ANNs are vulnerable to the Membership Inference Attack (MIA), but whether the same applies to SNNs is not explored. In this paper, we evaluate the membership privacy of SNNs by considering eight MIAs, seven of which are inspired by MIAs against ANNs. Our evaluation results show that SNNs are more vulnerable (maximum 10% higher in terms of balanced attack accuracy) than ANNs when both are trained with neuromorphic datasets (with time dimension). On the other hand, when training ANNs or SNNs with static datasets (without time dimension), the vulnerability depends on the dataset used. If we convert ANNs trained with static datasets to SNNs, the accuracy of MIAs drops (maximum 11.5% with a reduction of 7.6% on the test accuracy of the target model). Next, we explore the impact factors of MIAs on SNNs by conducting a hyperparameter study. Finally, we show that the basic data augmentation method for static data and two recent data augmentation methods for neuromorphic data can considerably (maximum reduction of 25.7%) decrease MIAs' performance on SNNs.

cs.CR↗

arXiv2024

EmoBack: Backdoor Attacks Against Speaker Identification Using Emotional Prosody

Speaker identification (SI) determines a speaker's identity based on their spoken utterances. Previous work indicates that SI deep neural networks (DNNs) are vulnerable to backdoor attacks. Backdoor attacks involve embedding hidden triggers in DNNs' training data, causing the DNN to produce incorrect output when these triggers are present during inference. This is the first work that explores SI DNNs' vulnerability to backdoor attacks using speakers' emotional prosody, resulting in dynamic, inconspicuous triggers. We conducted a parameter study using three different datasets and DNN architectures to determine the impact of emotions as backdoor triggers on the accuracy of SI systems. Additionally, we have explored the robustness of our attacks by applying defenses like pruning, STRIP-ViTA, and three popular preprocessing techniques: quantization, median filtering, and squeezing. Our findings show that the aforementioned models are prone to our attack, indicating that emotional triggers (sad and neutral prosody) can be effectively used to compromise the integrity of SI systems. However, the results of our pruning experiments suggest potential solutions for reinforcing the models against our attacks, decreasing the attack success rate up to 40%.

cs.CR↗

arXiv2024

Context is the Key: Backdoor Attacks for In-Context Learning with Vision Transformers

Due to the high cost of training, large model (LM) practitioners commonly use pretrained models downloaded from untrusted sources, which could lead to owning compromised models. In-context learning is the ability of LMs to perform multiple tasks depending on the prompt or context. This can enable new attacks, such as backdoor attacks with dynamic behavior depending on how models are prompted. In this paper, we leverage the ability of vision transformers (ViTs) to perform different tasks depending on the prompts. Then, through data poisoning, we investigate two new threats: i) task-specific backdoors where the attacker chooses a target task to attack, and only the selected task is compromised at test time under the presence of the trigger. At the same time, any other task is not affected, even if prompted with the trigger. We succeeded in attacking every tested model, achieving up to 89.90\% degradation on the target task. ii) We generalize the attack, allowing the backdoor to affect \emph{any} task, even tasks unseen during the training phase. Our attack was successful on every tested model, achieving a maximum of $13\times$ degradation. Finally, we investigate the robustness of prompts and fine-tuning as techniques for removing the backdoors from the model. We found that these methods fall short and, in the best case, reduce the degradation from 89.90\% to 73.46\%.

cs.CR↗

arXiv2024

Continuous Embedding Attacks via Clipped Inputs in Jailbreaking Large Language Models

Security concerns for large language models (LLMs) have recently escalated, focusing on thwarting jailbreaking attempts in discrete prompts. However, the exploration of jailbreak vulnerabilities arising from continuous embeddings has been limited, as prior approaches primarily involved appending discrete or continuous suffixes to inputs. Our study presents a novel channel for conducting direct attacks on LLM inputs, eliminating the need for suffix addition or specific questions provided that the desired output is predefined. We additionally observe that extensive iterations often lead to overfitting, characterized by repetition in the output. To counteract this, we propose a simple yet effective strategy named CLIP. Our experiments show that for an input length of 40 at iteration 1000, applying CLIP improves the ASR from 62% to 83%

cs.CR↗

arXiv2024

A Comprehensive Study of Jailbreak Attack versus Defense for Large Language Models

Large Language Models (LLMS) have increasingly become central to generating content with potential societal impacts. Notably, these models have demonstrated capabilities for generating content that could be deemed harmful. To mitigate these risks, researchers have adopted safety training techniques to align model outputs with societal values to curb the generation of malicious content. However, the phenomenon of "jailbreaking", where carefully crafted prompts elicit harmful responses from models, persists as a significant challenge. This research conducts a comprehensive analysis of existing studies on jailbreaking LLMs and their defense techniques. We meticulously investigate nine attack techniques and seven defense techniques applied across three distinct language models: Vicuna, LLama, and GPT-3.5 Turbo. We aim to evaluate the effectiveness of these attack and defense techniques. Our findings reveal that existing white-box attacks underperform compared to universal techniques and that including special tokens in the input significantly affects the likelihood of successful attacks. This research highlights the need to concentrate on the security facets of LLMs. Additionally, we contribute to the field by releasing our datasets and testing framework, aiming to foster further research into LLM security. We believe these contributions will facilitate the exploration of security measures within this domain.

cs.CR↗

arXiv2024

Let's Focus: Focused Backdoor Attack against Federated Transfer Learning

Federated Transfer Learning (FTL) is the most general variation of Federated Learning. According to this distributed paradigm, a feature learning pre-step is commonly carried out by only one party, typically the server, on publicly shared data. After that, the Federated Learning phase takes place to train a classifier collaboratively using the learned feature extractor. Each involved client contributes by locally training only the classification layers on a private training set. The peculiarity of an FTL scenario makes it hard to understand whether poisoning attacks can be developed to craft an effective backdoor. State-of-the-art attack strategies assume the possibility of shifting the model attention toward relevant features introduced by a forged trigger injected in the input data by some untrusted clients. Of course, this is not feasible in FTL, as the learned features are fixed once the server performs the pre-training step. Consequently, in this paper, we investigate this intriguing Federated Learning scenario to identify and exploit a vulnerability obtained by combining eXplainable AI (XAI) and dataset distillation. In particular, the proposed attack can be carried out by one of the clients during the Federated Learning phase of FTL by identifying the optimal local for the trigger through XAI and encapsulating compressed information of the backdoor class. Due to its behavior, we refer to our approach as a focused backdoor approach (FB-FTL for short) and test its performance by explicitly referencing an image classification scenario. With an average 80% attack success rate, obtained results show the effectiveness of our attack also against existing defenses for Federated Learning.

cs.LG↗

arXiv2024

Label Inference Attacks against Node-level Vertical Federated GNNs

Federated learning enables collaborative training of machine learning models by keeping the raw data of the involved workers private. Three of its main objectives are to improve the models' privacy, security, and scalability. Vertical Federated Learning (VFL) offers an efficient cross-silo setting where a few parties collaboratively train a model without sharing the same features. In such a scenario, classification labels are commonly considered sensitive information held exclusively by one (active) party, while other (passive) parties use only their local information. Recent works have uncovered important flaws of VFL, leading to possible label inference attacks under the assumption that the attacker has some, even limited, background knowledge on the relation between labels and data. In this work, we are the first (to the best of our knowledge) to investigate label inference attacks on VFL using a zero-background knowledge strategy. To formulate our proposal, we focus on Graph Neural Networks (GNNs) as a target model for the underlying VFL. In particular, we refer to node classification tasks, which are widely studied, and GNNs have shown promising results. Our proposed attack, BlindSage, provides impressive results in the experiments, achieving nearly 100% accuracy in most cases. Even when the attacker has no information about the used architecture or the number of classes, the accuracy remains above 90% in most instances. Finally, we observe that well-known defenses cannot mitigate our attack without affecting the model's performance on the main classification task.

cs.LG↗

arXiv2024

The Power of Bamboo: On the Post-Compromise Security for Searchable Symmetric Encryption

Dynamic searchable symmetric encryption (DSSE) enables users to delegate the keyword search over dynamically updated encrypted databases to an honest-but-curious server without losing keyword privacy. This paper studies a new and practical security risk to DSSE, namely, secret key compromise (e.g., a user's secret key is leaked or stolen), which threatens all the security guarantees offered by existing DSSE schemes. To address this open problem, we introduce the notion of searchable encryption with key-update (SEKU) that provides users with the option of non-interactive key updates. We further define the notion of post-compromise secure with respect to leakage functions to study whether DSSE schemes can still provide data security after the client's secret key is compromised. We demonstrate that post-compromise security is achievable with a proposed protocol called ``Bamboo". Interestingly, the leakage functions of Bamboo satisfy the requirements for both forward and backward security. We conduct a performance evaluation of Bamboo using a real-world dataset and compare its runtime efficiency with the existing forward-and-backward secure DSSE schemes. The result shows that Bamboo provides strong security with better or comparable performance.

cs.CR↗

arXiv2024

A Systematic Evaluation of Evolving Highly Nonlinear Boolean Functions in Odd Sizes

Boolean functions are mathematical objects used in diverse applications. Different applications also have different requirements, making the research on Boolean functions very active. In the last 30 years, evolutionary algorithms have been shown to be a strong option for evolving Boolean functions in different sizes and with different properties. Still, most of those works consider similar settings and provide results that are mostly interesting from the evolutionary algorithm's perspective. This work considers the problem of evolving highly nonlinear Boolean functions in odd sizes. While the problem formulation sounds simple, the problem is remarkably difficult, and the related work is extremely scarce. We consider three solutions encodings and four Boolean function sizes and run a detailed experimental analysis. Our results show that the problem is challenging, and finding optimal solutions is impossible except for the smallest tested size. However, once we added local search to the evolutionary algorithm, we managed to find a Boolean function in nine inputs with nonlinearity 241, which, to our knowledge, had never been accomplished before with evolutionary algorithms.

cs.NE↗

arXiv2024

Time-Distributed Backdoor Attacks on Federated Spiking Learning

This paper investigates the vulnerability of spiking neural networks (SNNs) and federated learning (FL) to backdoor attacks using neuromorphic data. Despite the efficiency of SNNs and the privacy advantages of FL, particularly in low-powered devices, we demonstrate that these systems are susceptible to such attacks. We first assess the viability of using FL with SNNs using neuromorphic data, showing its potential usage. Then, we evaluate the transferability of known FL attack methods to SNNs, finding that these lead to suboptimal attack performance. Therefore, we explore backdoor attacks involving single and multiple attackers to improve the attack performance. Our primary contribution is developing a novel attack strategy tailored to SNNs and FL, which distributes the backdoor trigger temporally and across malicious devices, enhancing the attack's effectiveness and stealthiness. In the best case, we achieve a 100 attack success rate, 0.13 MSE, and 98.9 SSIM. Moreover, we adapt and evaluate an existing defense against backdoor attacks, revealing its inadequacy in protecting SNNs. This study underscores the need for robust security measures in deploying SNNs and FL, particularly in the context of backdoor attacks.

cs.CR↗

arXiv2024

Sneaky Spikes: Uncovering Stealthy Backdoor Attacks in Spiking Neural Networks with Neuromorphic Data

Deep neural networks (DNNs) have demonstrated remarkable performance across various tasks, including image and speech recognition. However, maximizing the effectiveness of DNNs requires meticulous optimization of numerous hyperparameters and network parameters through training. Moreover, high-performance DNNs entail many parameters, which consume significant energy during training. In order to overcome these challenges, researchers have turned to spiking neural networks (SNNs), which offer enhanced energy efficiency and biologically plausible data processing capabilities, rendering them highly suitable for sensory data tasks, particularly in neuromorphic data. Despite their advantages, SNNs, like DNNs, are susceptible to various threats, including adversarial examples and backdoor attacks. Yet, the field of SNNs still needs to be explored in terms of understanding and countering these attacks. This paper delves into backdoor attacks in SNNs using neuromorphic datasets and diverse triggers. Specifically, we explore backdoor triggers within neuromorphic data that can manipulate their position and color, providing a broader scope of possibilities than conventional triggers in domains like images. We present various attack strategies, achieving an attack success rate of up to 100% while maintaining a negligible impact on clean accuracy. Furthermore, we assess these attacks' stealthiness, revealing that our most potent attacks possess significant stealth capabilities. Lastly, we adapt several state-of-the-art defenses from the image domain, evaluating their efficacy on neuromorphic data and uncovering instances where they fall short, leading to compromised performance.

cs.CR↗