Perception models are fundamental to autonomous vehicles, enabling critical tasks such as object detection, drivable area segmentation, and lane line segmentation. Recently, multitask perception models that unify multiple vision tasks into a single architecture have shown their promise in efficiency and improved performance. However, their robustness under adversarial conditions remains underexplored. In this work, we present ARMBench, a benchmark designed to evaluate the adversarial robustness of multitask perception models for autonomous driving. ARMBench integrates a suite of representative attack methods as well as multiple defense mechanisms. Through a case study on the YOLOP model using the BDD100K dataset, we investigate the impact of adversarial attacks on individual tasks, compare the robustness of multitask and single-task models, and assess the effectiveness of defense strategies. Our results reveal asymmetric vulnerabilities across tasks and highlight the need for task-aware evaluation and defense. We have open-sourced ARMBench at https://github.com/Yunge6666/YOLOP-adversarial-attack-and-defense to contribute to the research community by providing a reproducible, extensible foundation for advancing robust and trustworthy multitask perception in AV systems.

The advancement of machine learning systems depends on large-scale, high-quality datasets. However, corpora drawn from user-generated and proprietary domains are often laden with sensitive information, posing significant privacy, security, and compliance risks. Conventional anonymization methods, which focus on removing explicit identifiers, can degrade downstream performance and leave the more nuanced challenge of implicit privacy leakage unresolved. This form of leakage allows for sensitive attributes such as author identity, demographics, or personality to be inferred from writing style alone. To address this, we present a privacy-preserving text rewriting framework based on guided reinforcement learning. Our approach features a composite reward function that operates over disentangled semantic and stylistic representations to preserve utility while enforcing style convergence and author anonymity. Empirical validation demonstrates substantial improvements on implicit privacy metrics without sacrificing semantic fidelity, yielding a scalable, model-agnostic solution for privacy-preserving data generation in the age of Large Language Models.

Artificial Intelligence (AI) agents capable of autonomous learning and independent decision-making hold great promise for addressing complex challenges across various critical infrastructure domains, including transportation, energy systems, and manufacturing. However, the surge in the design and deployment of AI systems, driven by various stakeholders with distinct and unaligned objectives, introduces a crucial challenge: How can uncoordinated AI systems coexist and evolve harmoniously in shared environments without creating chaos or compromising safety? To address this, we advocate for a fundamental rethinking of existing multi-agent frameworks, such as multi-agent systems and game theory, which are largely limited to predefined rules and static objective structures. We posit that AI agents should be empowered to adjust their objectives dynamically, make compromises, form coalitions, and safely compete or cooperate through evolving relationships and social feedback. Through two case studies in critical infrastructure applications, we call for a shift toward the emergent, self-organizing, and context-aware nature of these multi-agentic AI systems.

Identifying IoT devices is crucial for network monitoring, security enforcement, and inventory tracking. However, most existing identification methods rely on deep packet inspection, which raises privacy concerns and adds computational complexity. Moreover, existing works overlook the impact of wireless channel dynamics on the accuracy of layer-2 features, thereby limiting their effectiveness in real-world scenarios. In this work, we define and use the latency of specific probe-response packet exchanges, referred to as "device latency," as the main feature for device identification. Additionally, we reveal the critical impact of wireless channel dynamics on the accuracy of device identification based on device latency features. Specifically, this work introduces "accumulation score" as a novel approach to capturing fine-grained channel dynamics and their impact on device latency when training machine learning models. We implement the proposed methods and measure the accuracy and overhead of device identification in real-world scenarios. The results confirm that by incorporating the accumulation score for balanced data collection and training machine learning algorithms, we achieve an F1 score of over 97% for device identification, even amidst wireless channel dynamics, a significant improvement over the 75% F1 score achieved by disregarding the impact of channel dynamics on data collection and device latency.

Abstract Using machine learning and deep learning methods to analyze text data from social media can effectively explore hidden emotional tendencies and evaluate the psychological state of social media account owners. However, the label noise caused by mislabeling may significantly influence the training and prediction results of traditional supervised models. To resolve this problem, this paper proposes a psychological evaluation method that incorporates a noisy label correction mechanism and designs an evaluation framework that consists of a primary classification model and a noisy label correction mechanism. Firstly, the social media text data are transformed into heterogeneous text graphs, and a classification model combining a pre-trained model with a graph neural network is constructed to extract semantic features and structural features, respectively. After that, the Gaussian mixture model is used to select the samples that are likely to be mislabeled. Then, soft labels are generated for them to enable noisy label correction without prior knowledge of the noise distribution information. Finally, the corrected and clean samples are composed into a new data set and re-input into the primary model for mental state classification. Results of experiments on three real data sets indicate that the proposed method outperforms current advanced models in classification accuracy and noise robustness under different noise ratio settings, and can efficiently explore the potential sentiment tendencies and users’ psychological states in social media text data.

The Diffusion model has a strong ability to generate wild images. However, the model can just generate inaccurate images with the guidance of text, which makes it very challenging to directly apply the text-guided generative model for virtual try-on scenarios. Taking images as guiding conditions of the diffusion model, this paper proposes a brand new personalized virtual try-on model (PE-VITON), which uses the two stages (shape control and texture guidance) to decouple the clothing attributes. Specifically, the proposed model adaptively matches the clothing to human body parts through the Shape Control Module (SCM) to mitigate the misalignment of the clothing and the human body parts. The semantic information of the input clothing is parsed by the Texture Guided Module (TGM), and the corresponding texture is generated by directional guidance. Therefore, this model can effectively solve the problems of weak reduction of clothing folds, poor generation effect under complex human posture, blurred edges of clothing, and unclear texture styles in traditional try-on methods. Meanwhile, the model can automatically enhance the generated clothing folds and textures according to the human posture, and improve the authenticity of the virtual try-on. In this paper, qualitative and quantitative experiments are carried out on high-resolution paired and unpaired datasets, the results show that the proposed model outperforms the state-of-the-art model.

The rapid development of intelligent transportation systems and connected vehicles has highlighted the need for secure and efficient key management systems (KMS). In this paper, we introduce VDKMS (Vehicular Decentralized Key Management System), a novel Decentralized Key Management System designed specifically as an infrastructure for Cellular Vehicular-to-Everything (V2X) networks, utilizing a blockchain-based approach. The proposed VDKMS addresses the challenges of secure communication, privacy preservation, and efficient key management in V2X scenarios. It integrates blockchain technology, Self-Sovereign Identity (SSI) principles, and Decentralized Identifiers (DIDs) to enable secure and trustworthy V2X applications among vehicles, infrastructures, and networks. We first provide a comprehensive overview of the system architecture, components, protocols, and workflows, covering aspects such as provisioning, registration, verification, and authorization. We then present a detailed performance evaluation, discussing the security properties and compatibility of the proposed solution, as well as a security analysis. Finally, we present potential applications in the vehicular ecosystem that can leverage the advantages of our approach.

The increasing popularity of social media promotes the proliferation of misinformation, especially in the communities of Chinese-speaking diasporas, which has caused significant negative societal impacts. In addition, most of the existing efforts on misinformation mitigation have focused on English and other western languages, which makes numerous overseas Chinese a very vulnerable population to online disinformation campaigns. In this paper, we present AMICA, an information retrieval system for alleviating misinformation for Chinese Americans. AMICA dynamically collects data from popular social media platforms for Chinese Americans, including WeChat, Twitter, YouTube, and Chinese forums. The data are stored and indexed in Elasticsearch to provide advanced search functionalities. Given a user query, the ranking of social media posts considers both topical relevance and the likelihood of being misinformation.

Depression is one of the leading factors in global disability and a top driver for suicides. Studies have shown that depression has an effect on language usage. In recent years, especially during the COVID pandemic, social media platforms have become the de facto platform for many individuals to self-disclose or discuss mental health issues like depression. This trend presents a unique opportunity for researchers and healthcare professionals to detect potential mental illnesses for early intervention or treatment by taking advantage of the recent advances in machine learning approaches. Existing depression detection methods on social media, however, suffer from two major limitations. First, these solutions heavily rely on the amount, quality, and type of user-posted content. Second, the overlooked social circle impact should be leveraged to enhance the prediction capabilities. In this paper, we propose a depression detection framework, MentalNet, based on heterogeneous graph convolution by capturing users’ interactions (replies, mentions, and quotetiveets) with their friends on social media and differentiating the intimacy of users’ social circles (e.g., family, friends, or acquaintances). Specifically, we formulate the problem of depression detection on social media as a graph classification problem by representing users’ social circles in the format of heterogeneous graphs. MentalNet embraces three modules, (1) extraction of ego-network node features, (2) construction of user interaction graphs, and (3) depression detection based on heterogeneous graph classification. The extensive experiments on Twitter data demonstrate that MentalNet consistently and significantly outperforms the state-of-the-art methods in terms of all the effectiveness metrics. Compared to the baseline methods, MentalNet is able to effectively predict early depression in Twitter users with up to 24% improvement on F1 score.

Internet of Things (IoT) technologies (e.g., power-efficient occupancy-based energy management systems) are increasingly deployed in commercial buildings to reduce building energy consumption. However, the sensors involved in such systems are rarely adopted in residential houses due to their relatively high costs and users’ privacy concerns. Low-cost and nonintrusive IoT sensors have been proposed for residential houses for use with machine-learning algorithms. Furthermore, such sensors may be triggered very infrequently due to their nonintrusive nature, and it can take several days/weeks to collect sufficient training data. There is a research gap in accurately detecting occupancy information in residential houses with limited training data. This article proposes a trust-based occupancy detection scheme, which achieves high detection accuracy based on limited training data collected by nonintrusive, low-cost sensors. First, rather than directly taking raw sensor data as inputs, the semantic meanings (i.e., human activity sequences) are extracted from the data based on the order of triggered sensors. Second, the extracted human activity sequences are fed into the proposed trust-based sequence matching scheme for further occupancy detection. Comprehensive experimental results show that, when compared to existing occupancy detection algorithms, the proposed scheme can reliably achieve higher accuracy, especially when only limited training data is available.

IoT devices have become popular targets for various network attacks due to their lack of industry-wide security standards. In this work, we focus on the classification of smart home IoT devices and defending them against Distributed Denial of Service (DDoS) attacks. The proposed framework protects smart homes by using VLAN-based network isolation. This architecture includes two VLANs: one with non-verified devices and the other with verified devices, both of which are managed by a SDN controller. Lightweight, stateless flow-based features, including ICMP, TCP and UDP protocol percentage, packet count and size, and IP diversity ratio, are proposed for efficient feature collection. Further analysis is performed to minimize training data to run on resource-constrained edge devices in smart home networks. Three popular machine learning models, including K-Nearest-Neighbors, Random Forest, and Support Vector Machines, are used to classify IoT devices and detect different DDoS attacks based on TCP-SYN, UDP, and ICMP. The system’s effectiveness and efficiency are evaluated by emulating a network consisting of an Open vSwitch, Faucet SDN controller, and flow traces of several IoT devices from two different testbeds. The proposed framework achieves an average accuracy of 97%in device classification and 98% in DDoS detection with average latency of 1.18 milliseconds.

With the rise of Internet of Things (IoT) devices, home network management and security are becoming complex. There is an urgent requirement to make smart home network management more efficient. This work proposes an SDN-based architecture to secure smart home networks through K-Nearest Neighbor (KNN) based device classifications and malicious traffic detection. The efficiency is enhanced by offloading the computation-intensive KNN model to a Field Programmable Gate Arrays (FPGA). Furthermore, we propose a custom KNN solution that exhibits the best performance on an FPGA compared with four alternative KNN instances (i.e., 78% faster than a parallel Bubble Sort-based implementation and 99% faster than three other sorting algorithms). Moreover, with 36,225 training samples, the proposed KNN solution classifies a test query with 95% accuracy in approximately 4 ms on an FPGA compared to 57 seconds on a CPU platform. This highlights the promise of FPGA-based platforms for edge computing applications in the smart home.

Internet of Things (IoT) has facilitated the prosperity of smart environments such as smart homes. Meanwhile, WiFi is a broadly used technology for the wireless connectivity of IoT devices. However, smart home IoT devices are often vulnerable to various security attacks. This article quantifies the impact of distributed denial of service (DDoS) and energy-oriented DDoS attacks (E-DDoS) on WiFi smart home devices and explores the underlying reasons from the perspective of attacker, victim device, and access point (AP). Compared to the existing work, which primarily focus on DDoS attacks launched by compromised IoT devices against servers, our work focuses on the connectivity and energy consumption of IoT devices when under attack. Our key findings are threefold. First, the minimum DDoS attack rate causing service disruptions varies significantly among different IoT smart home devices, and buffer overflow within the victim device is validated as critical. Second, the group key updating process of WiFi may facilitate DDoS attacks by causing faster victim disconnections. Third, a higher E-DDoS attack rate sent by the attacker may not necessarily lead to a victim's higher energy consumption. Our study reveals the communication protocols, attack rates, payload sizes, and victim devices' ports state as the vital factors to determine the energy consumption of victim devices. These findings facilitate a thorough understanding of IoT devices' potential vulnerabilities within a smart home environment and pave solid foundations for future studies on defense solutions.

Digital signature is a major component of transactions on Blockchain platforms, especially in enterprise Blockchain platforms, where multiple signatures from a set of peers need to be produced to endorse a transaction. However, such process is often complex and time-consuming. Multi-signature, which can improve transaction efficiency by having a set of signers cooperate to produce a joint signature, has attracted extensive attentions. In this work, we propose two multi-signature schemes, GMS and AGMS, which are proved to be more secure and efficient than state-of-the-art multi-signature schemes. Besides, we implement the proposed schemes in a real Enterprise Blockchain platform, Fabric. Experiment results show that the proposed AGMS scheme helps achieve the goal of high transaction efficiency, low storage complexity, as well as high robustness against rogue-key attacks and $k$ -sum problem attacks.

Activity recognition using ubiquitous wearable devices (e.g., smartphones, smartwatches and sport bracelets) can be applied to many application domains such as healthcare, smart environments, assisted living, human-computer interaction, surveillance etc. Most existing activity recognition approaches require users to provide each activity a sufficient amount of annotations (labels) in order to achieve acceptable performance and therefore often fail to scale to a large number of activities by recognizing new (emerging) activities. To tackle this limitation, the systems requiring limited training data are much desired. However, existing activity recognition solutions on limited training data rely heavily on low-level activity or attribute extraction and therefore suffer from two major limitations: (1) failing to work well when activities are highly similar to each other, such as jogging, running, and jumping front and back, and (2) leading to overall system performance degradation on recognizing existing activities with sufficient training data. In this paper, we introduce SMART, a unified semi-supervised framework for recognizing highly similar emerging activities without sacrificing the performance on recognizing existing activities. Extensive experiments on real-world data showed that compared to the state of the art, SMART yielded superior performance on recognizing emerging activities, especially highly similar emerging activities, while providing comparable performance on recognizing existing activities.

In this paper, we propose GraphSE\textsuperscript2, an encrypted graph database for online social network services to address massive data breaches. GraphSE\textsuperscript2 ~preserves the functionality of social search, a key enabler for quality social network services, where social search queries are conducted on a large-scale social graph and meanwhile perform set and computational operations on user-generated contents. To enable efficient privacy-preserving social search, GraphSE\textsuperscript2 ~provides an encrypted structural data model to facilitate parallel and encrypted graph data access. It is also designed to decompose complex social search queries into atomic operations and realise them via interchangeable protocols in a fast and scalable manner. We build GraphSE\textsuperscript2 ~with various queries supported in the Facebook graph search engine and implement a full-fledged prototype. Extensive evaluations on Azure Cloud demonstrate that GraphSE\textsuperscript2 ~is practical for querying a social graph with a million of users.

With the proliferation of IoT, securing the abundance of devices is critical. The current IoT and security landscapes lack empirical evidence on algorithms optimized for constrained devices. In this paper, we study the performance of various symmetric encryption algorithms on a Contiki-based IoT device. This paper provides encryption and decryption durations and energy consumption results on three symmetric encryption algorithm implementations of AES (tinyAES, B-Con's AES, and Contiki's own built-in AES), where we found algorithms specifically built for constrained devices fared much better than those not, optimized algorithms using about 0.16 the energy and the time to perform encryption and decryption.

The Transport Layer Security (TLS) protocol has been considered as a promising approach to secure Internet of Things (IoT) applications. The different cipher suites offered by the TLS protocol play an essential role in determining communication security level. Each cipher suite encompasses a set of cryptographic algorithms, which can vary in terms of their resource consumption and significantly influence the lifetime of IoT devices. Based on these considerations, in this paper, we present a comprehensive study of the widely used cryptographic algorithms by annotating their source codes and running empirical measurements on two state-of-the-art, low-power wireless IoT platforms. Specifically, we present fine-grained resource consumption of the building blocks of the handshake and record layer algorithms and formulate tree structures that present various possible combinations of ciphers as well as individual functions. Depending on the parameters, a path is selected and traversed to calculate the corresponding resource impact. Our studies enable IoT developers to change cipher suite parameters and immediately observe the resource costs. Besides, these findings offer guidelines for choosing the most appropriate cipher suites for different application scenarios.

Online social networks (OSNs) are providing a variety of applications for human users to interact with families, friends, and even strangers. One such application, the friend search engine, allows the general public to query individual users' friend lists and has been gaining popularity recently. However, without proper design, this application may mistakenly disclose users' private relationship information. Our previous work has proposed a privacy preservation solution that can effectively boost OSNs' sociability while protecting users' friendship privacy against attacks launched by individual malicious requestors. In this paper, we propose an advanced collusion attack, where a victim user's friendship privacy can be compromised through a series of carefully designed queries coordinately launched by multiple malicious requestors. The effect of the proposed collusion attack is validated through synthetic and real-world social network data sets. The in-depth research on the advanced collusion attacks will help us design a more robust and secure friend search engine on OSNs in the near future.

In this paper, we propose a novel general non-negative matrix factorization (general-NMF)-based digital watermarking scheme for copyright protection and integrity authentication of the image content. Specifically, the proposed general-NMF algorithm is able to factorize a matrix C ∈ R+s×tinto a basis matrix A ∈ R+m×nand a coefficient matrix B ∈ R+p×qby removing the dimension-matching constraints required by the conventional NMF, where s = m, n = p, and t = q. In particular, s = m · l/n, t = l/p · q, and the variable l is the least common multiple of n and p. Furthermore, the generator factor of the random matrix and n are used as the keys of the proposed digital watermarking scheme. Experimental results show that the proposed digital watermarking scheme can effectively resist various attacks and tampering.

The prosperity of online rating system makes it a popular place for malicious vendors to mislead public's online decisions, whereas the security related studies are lagging behind. In this paper, we develop a quantile regression model to investigate influential factors on online user choices and reveal that the promotion effect on products' market outcomes is determined by not only the attacker's manipulation power but also the specific property of the target product and the market self-exciting power. Inspired by these findings, we propose a novel iterative rating attack and validate its effectiveness through experiments.

With the big success of the mobile application (app) sales, attackers are also attracted by the potential profits in the app market. In this paper, we survey current app ranking schemes as well as existing app reputation manipulation schemes and raise some interesting while arguable questions. Based on an app installation data set collected from a university campus community, we quantitatively investigate the answers to two questions: (1) what is the impact of app reputation and download number on app sales and (2) will attackers make profits from the manipulation of app reputation or download number. Although the results may not be generalized to the global app market, they provide a new view point for further investigations.

With the rapid development of reputation systems in various online social networks, manipulations against such systems are evolving quickly. In this paper, we propose scheme TATA, the abbreviation of joint Temporal And Trust Analysis, which protects reputation systems from a new angle: the combination of time domain anomaly detection and Dempster–Shafer theory-based trust computation. Real user attack data collected from a cyber competition is used to construct the testing data set. Compared with two representative reputation schemes and our previous scheme, TATA achieves a significantly better performance in terms of identifying items under attack, detecting malicious users who insert dishonest ratings, and recovering reputation scores.

The quality of life of leg amputees can be improved dramatically by using a cyber physical system (CPS) that controls artificial legs based on neural signals representing amputees' intended movements. The key to the CPS is the neural-machine interface (NMI) that senses electromyographic (EMG) signals to make control decisions. This paper presents a design and implementation of a novel NMI using an embedded computer system to collect neural signals from a physical system - a leg amputee, provide adequate computational capability to interpret such signals, and make decisions to identify user's intent for prostheses control in real time. A new deciphering algorithm, composed of an EMG pattern classifier and a post-processing scheme, was developed to identify the user's intended lower limb movements. To deal with environmental uncertainty, a trust management mechanism was designed to handle unexpected sensor failures and signal disturbances. Integrating the neural deciphering algorithm with the trust management mechanism resulted in a highly accurate and reliable software system for neural control of artificial legs. The software was then embedded in a newly designed hardware platform based on an embedded microcontroller and a graphic processing unit (GPU) to form a complete NMI for real time testing. Real time experiments on a leg amputee subject and an able-bodied subject have been carried out to test the control accuracy of the new NMI. Our extensive experiments have shown promising results on both subjects, paving the way for clinical feasibility of neural controlled artificial legs.