Lei Pan

Jiwei Guan, Tianyu Ding, Longbing Cao et al.

Vision-language pretraining (VLP) with transformers has demonstrated exceptional performance across numerous multimodal tasks. However, the adversarial robustness of these models has not been thoroughly investigated. Existing multimodal attack methods have largely overlooked cross-modal interactions between visual and textual modalities, particularly in the context of cross-attention mechanisms. In this paper, we study the adversarial vulnerability of recent VLP transformers and design a novel Joint Multimodal Transformer Feature Attack (JMTFA) that concurrently introduces adversarial perturbations in both visual and textual modalities under white-box settings. JMTFA strategically targets attention relevance scores to disrupt important features within each modality, generating adversarial samples by fusing perturbations and leading to erroneous model predictions. Experimental results indicate that the proposed approach achieves high attack success rates on vision-language understanding and reasoning downstream tasks compared to existing baselines. Notably, our findings reveal that the textual modality significantly influences the complex fusion processes within VLP transformers. Moreover, we observe no apparent relationship between model size and adversarial robustness under our proposed attacks. These insights emphasize a new dimension of adversarial robustness and underscore potential risks in the reliable deployment of multimodal AI systems.

Da Ma, Gonghu Shang, Zhi Chen et al.

Instruction tuning improves the ability of large language models (LLMs) to follow diverse human instructions, but achieving strong performance on specific target tasks remains challenging. A critical bottleneck is selecting the most relevant data to maximize task-specific performance. Existing data selection approaches include unstable influence-based methods and more stable distribution alignment methods, the latter of which critically rely on the underlying sample representation. In practice, most distribution alignment methods, from shallow features (e.g., BM25) to neural embeddings (e.g., BGE, LLM2Vec), may fail to capture how the model internally processes samples. To bridge this gap, we adopt a model-centric strategy in which each sample is represented by its neuronal activation pattern in the model, directly reflecting internal computation. However, directly using raw neuron activations leads to spurious similarity between unrelated samples due to neuron polysemanticity, where a single neuron may respond to multiple, unrelated concepts. To address this, we employ sparse autoencoders to disentangle polysemantic activations into sparse, monosemantic representations, and introduce a dedicated similarity metric for this space to better identify task-relevant data. Comprehensive experiments across multiple instruction datasets, models, tasks, and selection ratios show that our approach consistently outperforms existing data selection baselines in both stability and task-specific performance.

Jiwei Guan, Lei Pan, Chen Wang et al.

There are increasing concerns about malicious attacks on autonomous vehicles. In particular, inaudible voice command attacks pose a significant threat as voice commands become available in autonomous driving systems. How to empirically defend against these inaudible attacks remains an open question. Previous research investigates utilizing deep learning-based multimodal fusion for defense, without considering the model uncertainty in trustworthiness. As deep learning has been applied to increasingly sensitive tasks, uncertainty measurement is crucial in helping improve model robustness, especially in mission-critical scenarios. In this paper, we propose the Multimodal Fusion Framework (MFF) as an intelligent security system to defend against inaudible voice command attacks. MFF fuses heterogeneous audio-vision modalities using VGG family neural networks and achieves the detection accuracy of 92.25% in the comparative fusion method empirical study. Additionally, extensive experiments on audio-vision tasks reveal the model's uncertainty. Using Expected Calibration Errors, we measure calibration errors and Monte-Carlo Dropout to estimate the predictive distribution for the proposed models. Our findings show empirically to train robust multimodal models, improve standard accuracy and provide a further step toward interpretability. Finally, we discuss the pros and cons of our approach and its applicability for Advanced Driver Assistance Systems.

Khondker Fariha Hossain, Sharif Amit Kamran, Alireza Tavakkoli et al.

Electrocardiogram (ECG) acquisition requires an automated system and analysis pipeline for understanding specific rhythm irregularities. Deep neural networks have become a popular technique for tracing ECG signals, outperforming human experts. Despite this, convolutional neural networks are susceptible to adversarial examples that can misclassify ECG signals and decrease the model's precision. Moreover, they do not generalize well on the out-of-distribution dataset. The GAN architecture has been employed in recent works to synthesize adversarial ECG signals to increase existing training data. However, they use a disjointed CNN-based classification architecture to detect arrhythmia. Till now, no versatile architecture has been proposed that can detect adversarial examples and classify arrhythmia simultaneously. To alleviate this, we propose a novel Conditional Generative Adversarial Network to simultaneously generate ECG signals for different categories and detect cardiac abnormalities. Moreover, the model is conditioned on class-specific ECG signals to synthesize realistic adversarial examples. Consequently, we compare our architecture and show how it outperforms other classification models in normal/abnormal ECG signal detection by benchmarking real world and adversarial signals.

Yuantian Miao, Chao Chen, Lei Pan et al.

Stealing attack against controlled information, along with the increasing number of information leakage incidents, has become an emerging cyber security threat in recent years. Due to the booming development and deployment of advanced analytics solutions, novel stealing attacks utilize machine learning (ML) algorithms to achieve high success rate and cause a lot of damage. Detecting and defending against such attacks is challenging and urgent so that governments, organizations, and individuals should attach great importance to the ML-based stealing attacks. This survey presents the recent advances in this new type of attack and corresponding countermeasures. The ML-based stealing attack is reviewed in perspectives of three categories of targeted controlled information, including controlled user activities, controlled ML model-related information, and controlled authentication information. Recent publications are summarized to generalize an overarching attack methodology and to derive the limitations and future directions of ML-based stealing attacks. Furthermore, countermeasures are proposed towards developing effective protections from three aspects -- detection, disruption, and isolation.

Shuo Huang, Zhuang Li, Lizhen Qu et al.

Semantic parsing maps natural language (NL) utterances into logical forms (LFs), which underpins many advanced NLP problems. Semantic parsers gain performance boosts with deep neural networks, but inherit vulnerabilities against adversarial examples. In this paper, we provide the empirical study on the robustness of semantic parsers in the presence of adversarial attacks. Formally, adversaries of semantic parsing are considered to be the perturbed utterance-LF pairs, whose utterances have exactly the same meanings as the original ones. A scalable methodology is proposed to construct robustness test sets based on existing benchmark corpora. Our results answered five research questions in measuring the sate-of-the-art parsers' performance on robustness test sets, and evaluating the effect of data augmentation.

Nicholas Patterson, Michael Hobbs, Frank Jiang et al.

Virtual worlds have become highly popular in recent years with reports of over a billion people accessing these environments and the virtual goods market growing to near 50 billion US dollars. An undesirable outcome to this popularity and market value is thriving criminal activity in these worlds. The most profitable cyber security problem in virtual worlds is named Virtual Property Theft. The aim of this study is to use an online survey to gain insight into how hackers (n=100) in these synthetic worlds conduct their criminal activity. This survey is the first to report an insight into the criminal mind of hackers (virtual thieves). Results showed a clear-cut profile of a virtual property thief, they appear to be mainly aged 20-24 years of age, live in the United States of America, while using virtual worlds for 5-7 hours a day. These and the other key results of this study will provide a pathway for designing an effective anti-theft framework capable of abolishing this cyber security issue.

Yuantian Miao, Zichan Ruan, Lei Pan et al.

Network traffic analytics technology is a cornerstone for cyber security systems. We demonstrate its use through three popular and contemporary cyber security applications in intrusion detection, malware analysis and botnet detection. However, automated traffic analytics faces the challenges raised by big traffic data. In terms of big data's three characteristics --- volume, variety and velocity, we review three state of the art techniques to mitigate the key challenges including real-time traffic classification, unknown traffic classification, and efficiency of classifiers. The new techniques using statistical features, unknown discovery and correlation analytics show promising potentials to deal with big traffic data. Readers are encouraged to devote to improving the performance and practicability of automatic traffic analytic in cyber security.