Person search is a challenging computer vision task that aims to simultaneously detect and re-identify individuals from uncropped gallery images. However, most existing approaches are limited by restricted receptive fields, leading to distorted local feature representations under occlusions or complex poses. Additionally, scale variations hinder model generalization in real-world scenarios. To address these limitations, we introduce a novel E-Bike Rider Search (EBRS) dataset, which comprises 27,501 images capturing 963 distinct IDs across 8 camera views at a large urban intersection in a Chinese city. Furthermore, we propose a Context-aware Dynamic Contrastive Learning (CDCL) framework that dynamically adjusts convolutional weights and performs hard sample mining based on contextual cues, thereby improving discriminative capability for both local details and global features. Extensive experiments show our method achieves state-of-the-art performance on CUHK-SYSU and PRW benchmarks, with competitive results on the challenging EBRS dataset, demonstrating its effectiveness.

Multi-modal object Re-identification (ReID) aims to retrieve individuals by leveraging complementary information from different modalities. Recent CLIP-based approaches show promising results, but they usually employ prompt-based or hybrid prompt-adapter tuning and still face the problems of heterogeneous domain gap, fine-grained identity discrimination and noise instance interference. To address these problems, we introduce a novel Parameter-Efficient Fine-Tuning framework with Bag-of-Adapters (PEFT-BoA) based on the pre-trained CLIP's vision encoder for multi-modal object ReID. Specifically, we first propose a Domain-specific Patch Adapter (DPA) designed to bridge the visual feature gap between pre-trained and fine-tuned models at the local patch level. Meanwhile, we propose a Task-specific Class Adapter (TCA) enhance the fine-grained identity discrimination ability by optimizing global class token. Finally, we propose an Instance-specific Fusion Adapter (IFA) dynamically selects and combines only the most useful features across different modalities for each instance. Our PEFT-BoA achieves the better performance on multi-modal object re-identification benchmarks, while maintaining fewer trainable parameters (6.62M) and a higher training throughput (246.2fps).

In location-based social networks, next point of interest (POI) recommendation predicts the next-visited POIs of users by mining their behavioral patterns. However, existing POI recommendation methods based on graph neural networks and attention mechanisms fail to adequately capture: 1) the local structural features influenced by the trajectories of other users (i.e., the relationships between POIs visited by users); and 2) the dynamic visitation preferences and channel relationships among POIs (i.e., interdependencies between contextual features). We propose a next-POI recommendation model based on graph structure and sequential pattern to address these limitations. The model generates graph representations that reflect the real-time preferences of users by extracting local structures from a global POI graph. In addition, we design a temporal-aware self-attentive graph convolutional network and a channel attention mechanism to capture the structural features of users’ sequential visitation tendencies and the latent feature-channel relationships between POIs, respectively. These components enhance the ability of the model to characterize dynamic user preferences and behavioral changes. The results demonstrate that our model outperforms baseline methods on three real-world datasets, validating its effectiveness in capturing both global and local information.

Accurate prediction of pedestrian trajectories is essential for safety in autonomous driving. Pedestrian movements are shaped by individual behaviors, interactions with surrounding agents, and implicit social relationships. The primary challenge lies in effectively modeling these intricate social interactions while integrating temporal dependencies. Existing approaches typically aggregate agent states but often fail to account for hidden social relationships and trajectory multimodality. To address these challenges, this article proposes the social temporal multihead attention gated recurrent unit (Social-TMAGRU), a model that integrates two key submodules: social multihead attention (S-MHA) and temporal social attention GRU (TSA-GRU). The S-MHA submodule captures implicit social relationships between pedestrians by modeling their latent interactions, while TSA-GRU handles temporal dependencies and evolving dynamics in pedestrian trajectories. Experimental evaluations of multiple public datasets demonstrate that the proposed model significantly outperforms existing state-of-the-art methods in predictive accuracy and computational efficiency. The model excels in complex and high-density environments, demonstrating robust performance in dynamically changing scenarios, making it highly suitable for real-world autonomous driving applications.

High-quality synthetic tabular data raises severe issues about potential misuse. Existing post-processing database fingerprinting schemes relying on either Primary Key (PK) or original database ( non- blind) cannot work well in PK-free table, and easily suffer from regeneration attack using Diffusion Models (DMs). Integrating watermarking into the generation process of tabular data is an effective solution, such as Tree-Ring (TR) and Tabular Watermarking (TabWak) for latent tabular DMs. But, these schemes only address copyright protection and lack strong liability guarantees (fingerprinting) in case of unauthorized redistribution against collusion attack. Thus, we propose a framework ofPK-free,Blind andCollusion-resistant syntheticTabularFingerprinting (PBC-TabFip) with DMs by readily incorporating with symmetric Tardos codes of arbitrary alphabet sizes. PBC-TabFip solves the problem of deleted or modified PK, and maintains a blind property which does not use original data to extract fingerprint. Based on PBC-TabFip, we actually propose binary TabFip and TabFip+, quaternary TabFip* and TabFip+* schemes, where “+” means using self-cloning. To identify malicious user, we use Bit Matching (BM) and Valid Bit Matching (VBM) mechanisms for our schemes. We theoretically deduce that TabFip with BM obtains the highest expected accuracy of fingerprint bit matching against random noises. In the case of binary fingerprint, we also demonstrate the fragility of TabFip without Tardos codes, resulting in high probability of detecting innocent, and stronger robustness of TabFip with Tardos codes against inversion-based collusion attack using different strategies. Experimentally, comparing our schemes with baselines on two datasets, we show that the quality of synthetic table achieved by TabFip is higher than that achieved by other methods (TR under proportion of same rows$\lesssim$10%); and TabFip with BM achieves higher average detecting rate of correct user against five single-handed post-editing attacks. In a practical scenario, we exhibit that TabFip with Tardos codes identifies at least one of the colluders with 100% probability and without detecting innocent against two types of collusion attack.

Identifying the hierarchical heavy hitters (HHHs), i.e., the frequent aggregated flows based on common IP prefixes, is a vital task in network traffic measurement and security. Existing methods typically employ dynamic trie structures to track numerous prefixes or utilize multiple separate sketch instances, one for each hierarchical level, to capture HHHs across different levels, while both approaches suffer from low memory efficiency and limited compatibility with programmable switches. In this paper, we introduce two novel HHH detection solutions, respectively, Hierarchical Heavy Detector (HHD) and the Compressed Hierarchical Heavy Detector (CHHD), to achieve high memory efficiency and enhanced hardware compatibility. The key idea of HHD is to design a shared bucket array structure to identify and record HHHs from all hierarchical levels, which avoids the memory wastage of maintaining separate sketches to achieve high memory efficiency and allows feasible deployment of both byte-hierarchy and bit-hierarchy HHH detection on programmable switches using minimal processing stage resources. Additionally, HHD utilizes a sampling-based update strategy to effectively balance packet processing speed and detection accuracy. Furthermore, we present the CHHD, which enhances HHH detection in bit hierarchies through a more compact cell structure, which allows for compressing several ancestor and descendant prefixes within a single cell, further boosting memory efficiency and accuracy. We have implemented HHD and CHHD on a P4-based programmable switch with limited switch resources. Experimental results based on real-world Internet traces demonstrate that HHD and CHHD outperform the state-of-the-art by achieving up to 56 percentage points higher detection precision and 2.6× higher throughput.

Data queries play a crucial role in the vehicular platoon, enabling vehicles to obtain traffic information about surrounding road conditions and personalized entertainment information services. However, data query requests from vehicles may expose the vehicle owner’s personal attributes and habits. Although several schemes can address these issues, they are incapable of countering collusion attacks between the coordinating vehicle and roadside units (RSUs). To solve this problem, in this paper we propose a privacy-preserving and collusion-resistant data query scheme, named PCDQ. Specifically, PCDQ uses the Paillier encryption and Chinese Remainder Theorem to protect the query privacy of vehicle owners, allowing the RSU to recover individual data query requests without associating them with the original vehicles. Next, the parameter update mechanism in PCDQ prevents the coordinating vehicle from obtaining the corresponding mapping information between vehicles and query parameters, thereby resisting collusion attacks between the coordinating vehicle and RSUs. In addition, identity-based signcryption is used to ensure secure parameter distribution among vehicles, and the batch verification enables efficient authentication of query requests. Detailed security proofs and analysis demonstrate that PCDQ satisfies multiple security properties, including resistance to collusion attacks and replay attacks, unlinkability, confidentiality, and authentication and data integrity. Experimental results show that, compared to existing solutions, PCDQ performs better in terms of computation overhead, communication overhead, and network performance.

Graph Neural Networks (GNNs) have been widely adopted in recommendation systems. When integrated into a federated learning framework, GNNs can enhance the model’s expressive capability. However, challenges arise in personalized representation and graph expansion due to the heterogeneity and locality of user data in federated recommendation systems. To address these challenges, we propose a federated recommendation model based on personalized attention and privacy-preserving dynamic graphs. The method first matches neighbor users for each selected client. Subsequently, it counts the interaction frequencies of items for both local and neighbor users to construct personalized weights, which captures the unique characteristics of different users. Additionally, we designs a method for constructing privacy-preserving dynamic graphs. In each round of federated training, the selected client adds pseudo-interaction items to its own interaction subgraph, perturbing the real interactions. After completing local training, the noisy interaction subgraph is incorporated into the global graph to capture higher-order connectivity information among users while safeguarding their interaction privacy. We conduct extensive experiments on three benchmark datasets, and the results demonstrate that the proposed PADG method achieves superior performance while effectively protecting privacy.

With the widespread adoption of smartphones and other mobile intelligent devices, Mobile Crowd Sensing (MCS) is widely used. Typically, the real locations of the workers and tasks must be submitted to the service platform to complete the task allocation. Therefore, the protection of location information has become a key factor in influencing user participation. To address the issue of location information leakage, we propose a location information protection method based on local differential privacy, which can protect the location privacy of workers and tasks while generating approximately accurate task allocation results. Firstly, we divide the region into$k$*$k$grids and merge girds with a similar dispersion to form clusters. Then, this paper utilizes the inverse sampling of the cumulative distribution function (CDF) of the flipped Huber distribution to generate a personalized noise location set for each cluster. Furthermore, the exponential mechanism is used to select the obfuscated location for each user. Finally, the platform selects workers based on the perturbed location to complete the task allocation. Theoretical analysis shows that our mechanism satisfies differential privacy and achieves an approximately accurate task allocation. Experimental results demonstrate that, compared to existing methods, this method exhibits superior performance across different datasets and effectively balances the utility of data and the protection of location privacy.

With the exponential growth of mobile devices, Mobile Crowdsensing (MCS) has emerged as a new paradigm for various types of tasks. However, most existing studies focus primarily on the utility of either buyers or sellers, with limited exploration of the task types for buyers and the user types for sellers. Therefore, this paper investigates the incentive mechanisms for different types of sellers under varying task types. In this study, we classify sellers into two groups: teams and individuals, and classify buyers' tasks into two categories: simple tasks and complex tasks. Considering the characteristics of different task types and seller groups, we design two auction schemes: the Two-Stage Auction Scheme for Simple Tasks (TDAS-S) and the Two-Stage Auction Scheme for Complex Tasks (TDAS-C). In the first stage, we design a seller auction model based on the Stackelberg game to ensure the maximization of the seller's utility. In the second stage, we design an auction model for buyers, utilizing a variant of the Vickrey Auction and marginal contribution theory to pay the selected sellers in both TDAS-S and TDAS-C, ensuring the maximization of the buyer's utility. We further prove that the proposed scheme satisfies properties such as individual rationality, truthfulness, and budget balance. Finally, through extensive experiments on real-world datasets, we demonstrate that our scheme effectively balances the utility conflicts between buyers and sellers, allowing each party to maximize their respective utilities.

With the rapid advancement of Mobile Crowdsensing (MCS) technology, its impact on daily life continues to grow, making it indispensable to modern society. However, user data collection and analysis pose significant privacy risks. Although existing privacy-preserving task allocation schemes incorporate some basic adjustments for personalized noise, they fail to dynamically adapt based on user distribution and primarily focus on single-objective constraints. To bridge this gap, we propose a Density-Aware Personalized Differential Privacy for Multi-Objective Task Allocation in Mobile Crowdsensing (PDPMTA) scheme that adaptively adjusts noise intensity based on the distribution density of user data in feature space, then obfuscates sensitive information using differential privacy techniques. PDPMTA introduces a hybrid optimization strategy combining Simulated Annealing (SA) with NSGA-II, where simulated annealing is periodically applied to population subsets to balance exploration and exploitation, achieving more effective convergence toward the Pareto front. Experimental results on the real-world datasets confirm the scheme's effectiveness in optimizing travel distance, platform costs, and cost-efficiency.

Federated learning, owing to its distributed characteristic, is particularly susceptible to data poisoning attacks. To address this issue, a wealth of defenses has been developed that aim to mitigate such attacks by limiting the adverse effects of malicious models on the aggregated result. However, most existing defense methods are designed only from the perspective of data filtering or model weighting, which leads to poor robustness and an exclusive reliance on a single server-side defense mechanism. To mitigate these vulnerabilities, we propose FedD2, a federated defense framework that integrates data augmentation and residual-based weighted aggregation. Firstly, FedD2 applies data augmentation techniques to regenerate and mix local datasets, enhancing data generalization. During the server aggregation stage, FedD2 leverages model residuals to detect abnormal updates and adaptively assign aggregation weights to local models, thereby reducing the influence of malicious clients. Comprehensive experiments performed on benchmark datasets demonstrate that FedD2 significantly enhances classification performance and exhibits strong robustness against data poisoning attacks.

Auctions have been widely applied in resource allocation due to their fairness and efficiency. For instance, platforms receive requests from service requesters and utilize auction theory to select suitable service providers. Existing studies typically assume that winners are determined based on bidders’ true valuations by allowing arbitrary transactions between requesters and providers, which can lead to serious valuation privacy leakage issues and limitations in application scenarios. Although some research has addressed these concerns using differential privacy techniques, they mostly rely on a trusted platform, and the introduction of noise results in utility loss, making them unsuitable for restricted auction contexts. To overcome these limitations, we propose a restricted double auction mechanism based on local differential privacy. Specifically, we extract the characteristics of the valuation data and constrain the noise addition probability density function based on the data features. Then we design a novel exponential selection mechanism that ensures that the relative positions of the obfuscated bids remain unchanged compared to the original valuations, while satisfying ε-local differential privacy. Furthermore, we develop an auction matching mechanism that maintains properties such as truthfulness under restricted allocation. The simulation results demonstrate that the proposed bid obfuscation mechanism ensures that the relative positions of the interfered bids remain unchanged while incurring low time overhead. Compared to existing mechanisms, our restrictive auction mechanism can generate greater social welfare while reducing the risk of valuation privacy leakage.

In the modern interconnected world, the popularization of networks and the rapid development of information technology led to the increasing security risks and threats in network systems. The existing intrusion detection system is constantly challenged by various malicious intrusion attacks. Machine learning algorithms have been widely used in intrusion detection. However, the model training requires the support of a sufficient high-quality samples, especially attack traffic data. Network intrusion detection datasets may not be shared between organizations due to data security and some privacy policy concerns. The federated learning framework is an optimal approach to address this issue, in which organizations collaborate to train a global model shared by multiple parties while keeping the data local to the client, guaranteeing the data privacy and security of all parties. However, there is a problem of class imbalance in the network traffic data owned by the organizations, which seriously affects the detection performance of the model and leads to a high consumption of model training time. Therefore, this study proposed a novel federated undersampling learning framework with Gini impurity, namely Fed-UGI. The framework is based on the hash-based block undersampling method to rebalance the client, which can solve the influence of imbalanced training data on the model detection performance and improve the model training efficiency. Moreover, the client weighted aggregation strategy based on Local Gini impurity can further optimize the effect of global model aggregation and reduce the impact of the dispersion degree and information difference in client data on model aggregation. In addition, extensive experiments on intrusion detection datasets show that compared to SOTA methods, the proposed Fed-UGI method has a good detection effect on the three metrics of F1-score, G-mean and AUC, the training time of the model is reduced by 51.76%-92.58%, especially in highly class imbalance situation.

Numerous trajectory data mining algorithms rely on rich map information for improved effectiveness, necessitating essential trajectory preprocessing steps, such as map matching. Current deep models are restricted to small road networks and cannot adapt to or learn large-scale, variable, and noisy trajectories. From a data-driven perspective, we effectively applied and developed a transformer function and proposed a transformer-based map matching (MMformer) for large-scale road networks. This model does not require maps and trajectories to be gridded, and directly runs on the vectors of trajectory points. The decoder module can learn, gather, and store the connections of road segments; therefore, the outputs of the decoder have a graph-like inductive bias. Trajectory point vector, including coming direction, enhances encoder and decoder performance. After experimenting with different embedding modules, the trajectory point vectors were embedded using a single linear layer. Extensive experiments demonstrated that MMformer can perform the map-matching task for large-scale road networks, and its map-matching accuracy on large-scale road networks is 10% higher than that of existing models on small road networks.

Most current session-based recommendations model session sequences solely based on the user's target behavior, ignoring the user's hidden preferences in auxiliary behaviors. Additionally, they use ordinary graphs to model one-to-one item correlations in the current session and fail to leverage other sessions to learn richer higher-order item correlations. To address these issues, a multi-behavior hypergraph contrastive learning model for session-based recommendations is proposed. This model represents all the sessions as global hypergraphs according to two types of behavior sequences. It employs contrastive learning to obtain global item embeddings, which are further aggregated to generate a global session representation that captures higher-order correlations of items from all session perspectives. A novel local heterogeneous hypergraph is designed for the current session to capture higher-order correlations between items with different behaviors in the current session, thus enhancing the local session representation. Additionally, a novel self-supervised signal is created by constructing a multi-behavior line graph, enhancing the global session representation. Finally, the local session representation, global session representation, and global item embedding are used to learn the predicted interaction probability of each item. Extensive experiments are conducted on three real datasets, and the results demonstrate that the proposed model significantly improves recommendation accuracy.

Previous advances in vehicle re-identification (ReID) are mostly reported under favorable lighting conditions, while cross-day-and-night performance is neglected, which greatly hinders the development of related traffic intelli-gence applications. This work instead develops a novel Day-Night Dual-domain Modulation (DNDM) vehicle re-identification framework for day-night cross-domain traf-fic scenarios. Specifically, a unique night-domain glare suppression module is provided to attenuate the headlight glare from raw nighttime vehicle images. To enhance ve-hicle features under low-light environments, we propose a dual-domain structure enhancement module in the feature extractor, which enhances geometric structures between ap-pearance features. To alleviate day-night domain discrep-ancies, we develop a cross-domain class awareness mod-ule that facilitates the interaction between appearance and structure features in both domains. In this work, we ad-dress the Day-Night cross-domain ReID (DN-ReID) prob-lem and provide a new cross-domain dataset named DN-Wild, including day and night images of 2,286 identities, giving in total 85,945 daytime images and 54,952 nighttime images. Furthermore, we also take into account the mat-ter of balance between day and night samples, and provide a dataset called DN-348. Exhaustive experiments demon-strate the robustness of the proposed framework in the DN-ReID problem. The code and benchmark are released at https://github.com/chenjingong/DN-ReID.

To address the problem of data sparsity in recommendation systems, various studies have used knowledge graphs as auxiliary information. These studies have employed multitask learning (MTL) to enhance recommendation performance. However, the shared information between tasks is not fully explored when using an MTL strategy for training both recommendation and knowledge graph-related tasks. Moreover, most studies cannot effectively model the knowledge sharing, consequently affecting recommendation performance. In response to these problems, we proposed a novel knowledge graph-based personalized multitask enhanced recommendation model. To explore the shared information between tasks, a relation attention mechanism was proposed to distinguish the relative importance of neighborhood information to the central entity. Additionally, we utilized a lightweight graph convolutional network to more effectively aggregate high-order neighborhood information from the knowledge graph. This approach improves the accuracy of neighborhood feature and ensures that more suitable shared information is obtained. Furthermore, we developed a linear interaction component to model knowledge sharing between recommendation and knowledge graph embedding tasks. This component allows for detailed feature interaction learning between items and entities, enhancing the shared feature representation, generalization capabilities, and overall performance of the recommendation system. The experimental results on three public datasets indicate that our model outperforms other benchmark models in CTR prediction and top-$\boldsymbol{K}$recommendation.

Traditional recommendation systems require users to upload local data to the server to generate recommendation results. In this process, users’ privacy is easy to disclose. Federated recommendations can solve the problem of local data privacy leakage, but the intermediate computing results are not protected. The existing work mainly protects the information in this process through encryption or disturbance schemes, which will result in complex calculations or low accuracy. Also, the two schemes only protect the rating data and process parameter information, but the existence information is not protected. Aiming at the above problems, this article proposes a federated matrix factorization based on secret sharing (FMFSS) to protect users’ privacy. The parameters are randomly divided into pieces, and then, the secret sharing technology is used to transmit private information between user–user and user–server, which does not introduce additional encryption cost and ensures value privacy, process privacy, and existence privacy. In addition, this article introduces the user–item interaction value, which is transmitted to the server with gradient information. In this way, the real gradient of the user cannot be inferred from the information received by the server from all parties, but the aggregated final average parameter information is real. The experimental comparison with the existing work and the analysis of computation time show that the proposed method can ensure the accuracy of model privacy recommendations without introducing additional encryption operations.

Vehicle re-identification (Re-ID) aims to retrieve vehicles across non-overlapping cameras. Most studies consider representation learning from single appearance information of the vehicle images. Some works adopt the spatio-temporal information to remove unreasonable vehicles to refine the results in the testing phase. However, they ignore the potential topological relations among cameras under the Closed Circuit Television (CCTV) camera systems in the training phase, which usually leads to suboptimal results due to the high intra-identity variations. To handle this problem, we propose a novel vehicle re-identification framework, which explicitly models the camera topological relations of all input images to aggregate neighbor images and thus acquires camera-independent representations. Specifically, we first construct a Camera Topology Graph (CTG) to elucidate the topological relations among cameras. It takes different cameras as nodes and constructs edges from four levels of the camera system, position, orientation, and individual. Then, we introduce a Camera Topology-based Graph Convolutional Network (CT-GCN), which suppresses irrelevant neighbor images and learns different camera representation functions. Finally, we propose a topological cross-entropy loss to obtain the more discriminative vehicle representations. The whole network is trained in an end-to-end manner. Extensive experiments on three benchmark datasets demonstrate the effectiveness of the proposed method against state-of-the-art vehicle Re-ID methods.

In the current matrix factorization recommendation approaches, the item and the user latent factor vectors are with the same dimension. Thus, the linear dot product is used as the interactive function between the user and the item to predict the ratings. However, the relationship between real users and items is not entirely linear and the existing recommendation model of matrix factorization faces the challenge of data sparsity. To this end, we propose a kernelized deep neural network recommendation model in this article. First, we encode the explicit user-item rating matrix in the form of column vectors and project them to higher dimensions to facilitate the simulation of nonlinear user-item interaction for enhancing the connection between users and items. Second, the algorithm of association rules is used to mine the implicit relation between users and items, rather than simple feature extraction of users or items, for improving the recommendation performance when the datasets are sparse. Third, through the autoencoder and kernelized network processing, the implicit data are connected with the explicit data by the multilayer perceptron network for iterative training instead of doing simple linear weighted summation. Finally, the predicted rating is output through the hidden layer. Extensive experiments were conducted on four public datasets in comparison with several existing well-known methods. The experimental results indicated that our proposed method has obtained improved performance in data sparsity and prediction accuracy.

Sequential recommendation models analyze users’ historical interactions to predict the next item they will en gage with. In order to better capture users’ dynamic interest preferences, most existing sequential recommendation models that introduce heterogeneous time intervals lead to increased model complexity, which raises computational costs and training difficulty. This is particularly evident in long sequential data, where the model need to handle a large variety of different time intervals. Additionally, accurately modeling the impact of long time intervals on user behavior remains a significant challenge. To address these issues, we propose a lightweight sequential recommendation algorithm with time interval awareness augmen tation (TALSAN). This model introduces a novel uniform data augmentation operator to improve the distribution of original data samples and employs a time-aware self-attention layer to model user interactions, maintaining the continuity of the original sequence. By integrating temporal context with posi tional features, TALSAN constructs a streamlined self-attention network for predicting user behavior. Comparative testing on datasets such as ML-100K, ML-1M, Amazon Beauty, Amazon Toys, and Amazon Fashion demonstrates the model’s superiority over existing baselines. Our results confirm that TALSAN not only mitigates cold start issues but also enhances the ability to learn user preferences, leading to improved prediction accuracy.

The recommendation system (RS) predicts user ratings by collecting user information, but the users’ private information may be exposed in this process. Thus, it is crucial to achieving a balance between recommendation performance and privacy-preserving of RSs. Aiming to solve the above problem, this article proposes a novel matrix factorization (MF) algorithm. The algorithm predicts the user rating through a linear weighting of global average rating, item average rating, user average rating, and MF, which improves the prediction accuracy. Then, based on the above algorithm, this article proposes a MF RS for preserving user privacy by using local differential privacy technology. In this algorithm, the rating data are normalized on the user side to reduce global sensitivity. Then, Laplace noise is added to sensitive data before it is sent to the aggregator. Finally, based on the disturbed data, rating prediction is realized by using the MF algorithm. The proposed method is compared with five well-known recommendation methods on four public datasets. The experimental results show that the proposed algorithm achieves better recommendation performance at the same privacy-preserving level.

Deep learning has been used in many computer-vision-based industrial Internet of Things applications. However, deep neural networks are vulnerable to adversarial examples that have been crafted specifically to fool a system while being imperceptible to humans. In this article, we propose a consensus defense (Cons-Def) method to defend against adversarial attacks. Cons-Def implements classification and detection based on the consensus of the classifications of the augmented examples, which are generated based on an individually implemented intensity exchange on the red, green, and blue components of the input image. We train a CNN using augmented examples together with their original examples. For the test image to be assigned to a specific class, the class occurrence of the classifications on its augmented images should be the maximum and reach a defined threshold. Otherwise, it is detected as an adversarial example. The comparison experiments are implemented on MNIST, CIFAR-10, and ImageNet. The average defense success rate (DSR) against white-box attacks on the test sets of the three datasets is 80.3%. The average DSR against black-box attacks on CIFAR-10 is 91.4%. The average classification accuracies of Cons-Def on benign examples of the three datasets are 98.0%, 78.3%, and 66.1%. The experimental results show that Cons-Def shows a high classification performance on benign examples and is robust against white-box and black-box adversarial attacks.

In view of the lack of research regarding characteristic factors (e.g., service cost, quality, etc.) in trust models for cloud environments, as well as the lack of negotiation and monitoring mechanisms, a trust model based on characteristic factors and service level agreements (SLAs) is proposed in this article. First, on the basis of comprehensive trust and self-recommended trust, we introduce the cost deviation trust and service quality coefficient, to simultaneously consider the influence of common and characteristic factors upon trust evaluations and thereby improve the accuracy of trust evaluations. Second, we establish a negotiation and monitoring mechanism whereby both parties sign an SLA before the trade and require SLA agent monitoring services to improve the accuracy of the service cost and quality evaluation and the efficiency of malicious entity identification. Finally, using the agreement quality, experience quality, and monitoring quality, we more accurately judge the trade result and update the recognition degree (which plays a key role in trust evaluation), to further improve the accuracy of trust evaluation and thereby the trade success rate. The results of experiments conducted upon real datasets show that our model can effectively resist spoofing, coordination, and defamation attacks from malicious entities, and it offers a high trade success rate.

Abstract Spatial query is an important supporting technology in the Internet of Things (IoT) and location‐based services (LBS). The k‐nearest neighbor query is widely used for spatial queries. However, user location privacy may be leaked in the query. In addition, some users are malicious or uncooperative. With the objective of overcoming these problems, a k‐nearest neighbor query method based on trust and location privacy protection is proposed. First, we employ a new K‐anonymity method based on cooperation to protect a query user's location privacy. In this method, the query user constructs an anonymous group by introducing a trust mechanism to incentivize cooperation among users. Then, according to the different radii of the selection area set by the query user, agent users with higher reputation values who send query requests for the query user are selected. Finally, the agent users obtain the query results from the LBS server and forward them to the query user, and the query user screens the results according to his or her real location. The experiments show that our method can effectively stimulate users to cooperate, better exclude malicious users, and improve the accuracy of the query results while protecting the privacy of the query user.

Overlapping community detection is essential for revealing the hidden structure of complex networks. In this work, we present an overlapping community detection algorithm that selects community centers adaptively based on density peaks. The proposed algorithm, called the density-peak-based overlapping community detection (DPOCD) algorithm, defines point link strength and edge link strength to construct distance matrix. Unlike the density peaks clustering algorithm, by which cluster centers are selected manually, the DPOCD algorithm uses the linear fitting method to select community centers. To evaluate the feasibility of the presented algorithm, we compared it with other advanced methods on artificial synthetic network and real complex network datasets. The experimental results demonstrate that our method achieves excellent performance in large-scale complex networks and the robustness of the algorithm.

Accurately mapping the raw global position system (GPS) trajectories to the road network is the basis for studying the application of trajectory data. This study proposes a novel off-line map matching algorithm based on road network topology, to address the problems of low execution efficiency and poor matching accuracy of selective look-ahead map matching (SLAMM) algorithm. First, the noise points of the trajectory data are removed by data preprocessing. Second, the algorithm searches for critical samples in the trajectory data and segments the data accordingly. Then, the adjacent road segments around the transition node corresponding to the critical sample are selected as candidate arcs. Finally, the segmented trajectory data are matched to the road network by constructing an error ellipse. The algorithm fully considers the topology of the road network and the characteristics of high-frequency trajectory data. The experimental results, using Beijing trajectory data to perform matching on an actual road network environment, show that the proposed algorithm is more efficient and robust than other map matching algorithms for high-frequency trajectories.

Abstract Aiming at overcoming problems such as malicious entities and trust crises in cloud environments, a trust management model based on mutual trust and a reward‐with‐punishment mechanism is proposed in this paper. First, according to reputation values of entities and trust relationships among entities, we calculate the comprehensive trust values of a service request user to several candidate service providers and probe these providers with high comprehensive trust values. Second, we select a trade provider according to mutual trust. Finally, after the trade, we update the reputation values and other data of the request user, trade provider, and recommenders based on the final trade result to reward or punish them to various degrees, thus ultimately reducing malicious or dishonest behavior. The experimental results show that our model can effectively identify malicious entities to increase the trade success rate.

Residents’ trajectory data denote their instantaneous locations along their movements. Mobility research that applies trajectory mining techniques to identify the transportation modes of these movements can inform urban transportation planning. Herein, we propose a five-step approach with information entropy and a multi-layer neural network to identify transportation modes from trajectory data. First, this approach extracts the motion features at each time-stamped location based on foundation geospatial data and spatiotemporal trajectory data, including the speed, acceleration, change of direction, rate of change in direction, and distance from each basic transportation facility. The second step uses information entropy to identify the features that play key roles in identifying transportation modes. The third step weighs each attribute in the feature vector consisting of the selected features and normalizes it to prepare it as input data. The fourth step constructs, trains, and tests a multi-layer neural network with seven-fold cross-validation. The final step includes a post-processing method to optimize the identification result. We use F-measure metric to evaluate the performance. Experimental results on a real trajectory dataset show that the proposed approach can identify the transportation mode at each time-stamped location and outperforms existing transportation-mode identification methods in terms of accuracy and stability.

Trajectory data may include the user’s occupation, medical records, and other similar information. However, attackers can use specific background knowledge to analyze published trajectory data and access a user’s private information. Different users have different requirements regarding the anonymity of sensitive information. To satisfy personalized privacy protection requirements and minimize data loss, we propose a novel trajectory privacy preservation method based on sensitive attribute generalization and trajectory perturbation. The proposed method can prevent an attacker who has a large amount of background knowledge and has exchanged information with other attackers from stealing private user information. First, a trajectory dataset is clustered and frequent patterns are mined according to the clustering results. Thereafter, the sensitive attributes found within the frequent patterns are generalized according to the user requirements. Finally, the trajectory locations are perturbed to achieve trajectory privacy protection. The results of theoretical analyses and experimental evaluations demonstrate the effectiveness of the proposed method in preserving personalized privacy in published trajectory data.

With the rapid development of portable mobile devices, mobile crowd sensing systems (MCS) have been widely studied. However, the sensing data provided by participants in MCS applications is always unreliable, which affects the service quality of the system, and the truth discovery technology can effectively obtain true values from the data provided by multiple users. At the same time, privacy leaks also restrict users’ enthusiasm for participating in the MCS. Based on this, our paper proposes a secure truth discovery for data aggregation in crowd sensing systems, STDDA, which iteratively calculates user weights and true values to obtain real object data. In order to protect the privacy of data, STDDA divides users into several clusters, and users in the clusters ensure the privacy of data by adding secret random numbers to the perceived data. At the same time, the cluster head node uses the secure sum protocol to obtain the aggregation result of the sense data and uploads it to the server so that the server cannot obtain the sense data and weight of individual users, further ensuring the privacy of the user’s sense data and weight. In addition, using the truth discovery method, STDDA provides corresponding processing mechanisms for users’ dynamic joining and exiting, which enhances the robustness of the system. Experimental results show that STDDA has the characteristics of high accuracy, low communication, and high security.

In this poster, we propose a novel, effective tuple-based anonymization technique for categorical data over the Internet. By utilizing a new structure, called Candidate Encoding Sequence with Frequency, and a set of new rules for generating such a sequence for each domain value of the categorical data, we can effectively solve the key limitation of the existing methods. Our experimental results demonstrate the superiority of our method against the existing method in terms of the strength of privacy protection.

With the widespread adoption of cloud computing, large-scale online applications composed of services have been deployed in many critical areas. In order to ensure the performance of cloud applications, Quality of Service (QoS) is a key indicator commonly used for service selection and adaptation. Previous studies have proposed collaborative QoS prediction approaches to estimate personalized QoS values. However, collaborative QoS prediction encounters privacy problems in practice. As a result, privacy threat has become a key challenge to make QoS prediction approaches practical. In this paper, we proposed a privacy-preserving QoS prediction approach employing federated learning techniques to tackle this grand challenge. We further improve the prediction efficiency by reducing system overhead and make the federated privacy-preserving QoS prediction approach feasible. The proposed approach is evaluated on a large-scale real-world QoS dataset, and the experimental results confirm its effectiveness and efficiency.

With the widespread adoption of the fifth-generation (5G) cellular network and Mobile Edge Computing (MEC), numerous Internet of Things (IoT) applications are emerging in many critical areas. IoT applications are typically running on mobile devices to provide real-time interaction with users by connecting with smart IoT devices and remote cloud services. In order to ensure the performance of IoT applications, Quality of Service (QoS) is commonly used as a key metric for the selection and adaptation of high-quality services at runtime. Collaborative QoS prediction methods have been proposed in the literature to predict personalized QoS values, enabling QoS-based selection and adaptation. However, privacy issues in collaborative QoS prediction discourage users from collaborating by sharing data in practice. Furthermore, there are untrusted users in the unreliable MEC environment, which makes the prediction encounter serious reliability issues. As a result, privacy and reliability issues have become key challenges to make QoS prediction approaches feasible. In this paper, we proposed a credible and privacypreserving QoS prediction approach by leveraging federated learning techniques and developing reputation mechanisms to address this critical challenge. We evaluate the method on a large- scale real QoS dataset and the experimental results demonstrate the effectiveness and efficiency of the method.

Summary Elevator scheduling algorithms generally denote methods used to calculate how to use the elevator. These algorithms can distribute elevators to various floors of a building, thereby achieving efficient transportation. From the perspective of the elevator scheduling problem, we address the load‐balancing problem for software‐defined networking (SDN) architecture and propose a load‐balancing method based on the elevator scheduling algorithm, LBBESA. We take advantage of the flexibility of the SDN architecture, obtain the real‐time load of the server through real‐time statistical analyses of the SDN switch port traffic by the controller, and combine this with the idea of regional elevator allocation to coordinate the connection of the client's requests and realize the load balancing of each server in the cluster. Simulation experiments show that, compared with the round‐robin algorithm, LBBESA is more effective in the load balancing of the server pool and can improve the throughput of the server pool to a certain extent. In addition, our scheme is easy to implement and has high scalability.

Summary Aiming at preventing the privacy disclosure of sensitive information, issues related to privacy protection in cloud computing have attracted the interest of researchers. To protect the privacy of users during clustering in a cloud computing environment, we present a privacy‐preserving density peak clustering (PPDPC) algorithm that neither discloses personal privacy information nor leaks the cluster centers. Our scheme contains two steps of density peak clustering: First, a cloud service provider calculates the cluster centers without knowing each participant's private data and without disclosing any cluster center information to the other participants, and second, participant allocation is secure and every participant is prevented from identifying the other members of the same cluster. Security analysis and comparison experiments show that the proposed PPDPC algorithm not only obtains good accuracy with respect to density peak clustering but also resists collusion attacks even if the cloud service provider is collaborating with all except one participant. Both theoretical analysis and experimental results confirm the security and accuracy of our method.

Big trajectory data feature analysis for mobile networks is a popular big data analysis task. Due to the large coverage and complexity of the mobile networks, it is difficult to define and detect anomalies in urban motion behavior. Some existing methods are not suitable for the detection of abnormal urban vehicle trajectories because they use the limited single detection techniques, such as determining the common patterns. In this study, we propose a framework for urban trajectory modeling and anomaly detection. Our framework takes into account the fact that anomalous behavior manifests the overall shape of unusual locations and trajectories in the spatial domain as well as the way these locations appear. Therefore, this study determines the peripheral features required for anomaly detection, including spatial location, sequence, and behavioral features. Then, we explore sports behaviors from the three types of features and build a taxi trajectory model for anomaly detection. Anomaly detection, including sports behaviors, are (i) detour behavior detection using an algorithm for global router anomaly detection of trajectories having a pair of same starting and ending points; this method is based on the isolation forest algorithm; (ii) local speed anomaly detection based on the DBSCAN algorithm; and (iii) local shape anomaly detection based on the local outlier factor algorithm. Using a real-life dataset, we demonstrate the effectiveness of our methods in detecting outliers. Furthermore, experiments show that the proposed algorithms perform better than the classical algorithm in terms of high accuracy and recall rate; thus, the proposed methods can accurately detect drivers’ abnormal behavior.

The issue of privacy preservation is receiving more and more attention when publishing trajectory data. In this paper, we study the challenges of published trajectory data anonymization. Most existing anonymization methods directly delete the trajectories or locations violating specific constraints , it is likely to cause a large loss of information. To address the problem, this paper proposes a trajectory privacy preservation method based on 3D-Grid partition in order to reduce information loss in the process of trajectory anonymization. This method first divides the trajectory region into several spatio-temporal units (denoted as 3D-cells), and then conducts location exchange or suppression in each spatio-temporal unit. Based on the trajectory data partition, within each 3D-cell, the proposed method exchanges locations among trajectories or removes very few locations of some sub-trajectories which do not meet the conditions rather than the whole trajectory. Our method considers three scenarios of trajectory distribution and measures trajectory similarity based on time, orientation, spatial locations and other features of trajectory. After the reconstruction of the related anonymous sub-trajectories, an anonymized trajectory dataset is obtained. Theoretical analysis and experimental results show that, compared to other methods, the proposed algorithm effectively preserves trajectory data privacy and improves the anonymous results of trajectory data in terms of accuracy and availability.

Existing location-privacy-preserving methods primarily focus on solving the problem of location-privacy preservation in the global space. This not only increases the response time of the location service, it also degrades the data quality. In this paper, a k-anonymity algorithm based on locality-se nsitive hashing is proposed to solve the problem of location-privacy preservation in the subspace. In the proposed algorithm, higher efficiency and higher quality of service are achieved by applying a bottom-up grid-search method. Further, reasonable division is obtained based on locality-sensitive hashing by retaining position characteristics. The results of experiments conducted to evaluate the proposed algorithm indicate that the proposed algorithm provides a smaller anonymous spatial region, higher data quality, and lower time cost than methods with no subspace.

The traditional trajectory privacy protection algorithm approaches the task as a single-layer problem. Taking a perspective in harmony with an approach more characteristic of human thinking, in which complex problems are solved hierarchically, we propose a two-level hierarchical granularity model f or this problem. The first level of the proposed model is a coarse-grained layer, in which the original dataset is divided into groups. The second level is a fine-grained layer, where problems are solved in each group instead of on the original dataset, which reduces complexity and computation while improving efficiency. On the basis of this hierarchical model, we propose the interpolation trajectory-anonymous privacy protection algorithm with temporal and spatial granularity constraints. In addition, we propose interpolation-based modified Hausdorff distance on adjacent segment (IMHD_AS), which provides a smaller clustering area and better data utility than the traditional Euclidean distance, as the trajectory similarity criterion for clustering within each group. Further, we theoretically prove that the proposed algorithm outperforms the traditional algorithm in terms of data distortion and anonymity cost and verify its efficacy experimentally. Compared with the classic anonymity algorithm, the maximum information loss and the anonymity cost are reduced by up to 21.04% and 28.32%, respectively.

Forecasting taxi demand is of great significance to the intelligent transportation systems in a smart city. Traditional demand prediction methods mostly considered about inter-regional traffic, events, activities, and weather, while they overlooked the influence of other travel modes, such as metro. In this paper, we propose a Deep Taxi-Metro Spatial-Temporal Network framework, namely TMST-Net, to model the spatiotemporal relationships between the taxi demand and the metro crowd flows. In detail, we apply residual neural networks to model temporal (current, day, and week) properties of the taxi demand in each area. For each feature, we apply residual convolutional units to handle the spatial properties of taxi demand. Likewise, we apply the same method to model the metro crowd flows. TMST-Net learns to assign different weights between taxi and metro by aggregating the output of the three residual neural networks and the external factors to forecast the final taxi demand for each area in the next timestamp. Experimental results on real taxi trajectory and the automatic fare collection (AFC) data in Shanghai show that our approach outperforms the state-of-the-art methods.

With the proliferation of various next generation wireless communication technologies, the smartphone has entered into our daily lives. Meanwhile, the excessive use of smartphone is seriously affecting our normal work and rest. In particular, the inappropriate way of using smartphone, especially when playing smartphone on bed in a lie-down posture, may do potential harm to our health. Thus, we propose a system called PosAla to encourage the smartphone users to use their phones in a correct posture. The proposed system adopts the supervised classifier to identify the harmful behavior, which is in a lie-down posture while using the smartphone, from a large number of sensor data generated during daily use of smartphone. Eight classification algorithms were tested in the experiments, and a large number of experiments have been conducted in terms of segment length and attribute selection. The extensive experimental results show that the recognition accuracy can reach 99.52%.

Data needs to be released to the relevant decision makers and researchers. Privacy protection should be carried out first because it contains personal sensitive information. The k-anonymity algorithm is an important privacy protection algorithm, and partitioning is one of its key methods. To reduce the computational complexity and low speed of existing privacy-preserving algorithms for high-dimensional data publishing, a probabilistic optimal projection partition k-dimensional (KD)-tree k-anonymity algorithm is proposed. First, some attribute dimensions are probabilistically selected from the global domain. Then, for these dimensions, the partition coefficient is calculated and the optimal partition point is determined. Furthermore, an improved KD-tree structure is introduced in which a node is a collection rather than a data point. The proposed KD-tree node is divided into left and right child nodes by the hyper-plane passing through the dividing point and perpendicular to the optimal dimension. The proposed algorithm is validated by a theoretical analysis and comparison experiments. The results show that the proposed algorithm can reduce the average generalization range by 11% to 22% compared to traditional k-anonymity. This enables better division and better dataset availability. Moreover, the runtime is reduced by 8% to 32% compared to globally optimal projection partitioning k-anonymity.

With the advent of virtualization technologies, cloud computing resource allocation issue plays an important role. However, the existing studies have not fully considered the heterogeneous demands from different cloud tenants. To tackle this, we design a more flexible cloud resource allocation mechanism which can maximize the profit of the cloud provider and support three general types of resource requirements from the cloud tenants. In this work, the jobs from tenants will bid for the usage of VMs in 3 types: 1) fixed time intervals, 2) time window intervals and 3) Time window slice intervals. We proved that the proposed approximation allocation mechanism has an approximation factor which approaches 1.58 when cmcloses to infinity.

Protecting users' privacy when transmitting a large amount of data over the Internet is becoming increasingly important nowadays. In this paper, we focus on the streaming categorical information and propose a novel anonymization technique for providing a strong privacy protection to safeguard against privacy disclosure and information tampering. Our technique utilizes an innovative two-phase anonymization approach which is very easy to implement, highly efficient in terms of speed and communication and is robust against possible tampering from adversaries. Extensive experimental evaluation that is conducted demonstrates that our technique is very efficient and more robust than the existing method.

Outlier detection is an interesting issue in data mining and machine learning. In this paper, to detect outliers, an information-entropy-based k-nearest neighborhood relevant outlier factor algorithm is proposed that is combined with Shannon information theory and the triangle pruning strategy. The algorithm accounts for the data points whose k-nearest neighbors are distributed on the edge of the range within the designated radius. In particular, the neighborhood influence on each point is considered to address the problem of information concealment and submergence. Information entropy is used to calculate the weights to distinguish the importance of each attribute. Then, based on the attribute weights, the improved pruning strategy reduces the computational complexity of the subsequent procedures by removing some inliers and obtaining the outlier candidate dataset. Finally, according to the weighted distance between the objects in the candidate dataset and those in the original dataset, the algorithm calculates the dissimilarity between each object and its k-nearest neighbors. The data points with the top $r$ dissimilarity are regarded as the outliers. Experimental results show that, compared to existing methods, the proposed approach improves pruning and detection rates while maintaining the coverage rate.

The use of intelligent technologies for providing useful recommendations to patients suffering chronic diseases may play a positive role in improving the general life quality of patients and help reduce the workload and cost involved in their daily healthcare. The objective of this study is to deve lop an intelligent recommender system based on predictive analysis for advising patients in the telehealth environment concerning whether they need to take the body test one day in advance by analyzing medical measurements of a patient for the past k days. The proposed algorithms supporting the recommender system have been validated using a time series telehealth data recorded from heart disease patients which were collected from May to January 2012, from our industry collaborator Tunstall. The experimental results show that the proposed system yields satisfactory recommendation accuracy and offer a promising way for saving the workload for patients to conduct body tests every day. This study highlights the possible usefulness of the computerized analysis of time series telehealth data in providing appropriate recommendations to patients suffering chronic diseases such as heart diseases patients.

Protecting users' privacy when transmitting a large amount of data over the Internet is becoming increasingly important nowadays. In this paper, we focus on the streaming choice-based information and propose a novel anonymization technique for providing a strong privacy protection to safeguard against privacy disclosure and information tampering. Our technique utilizes an innovative two-phase encoding-and-decoding approach which is very easy to implement, highly efficient in terms of speed and communication, and is robust against possible tampering from adversaries. The experimental evaluation demonstrates the promising performance of our technique.

String matching is a basic problem of string operation, and privacy-preserving string matching, as a special case of secure multi-party computation, has broad applications in auction, bidding and some other commercial areas. In this paper, some protocols are proposed to solve this private matching problem, the security and correctness are analyzed respectively, and the actual efficiency is tested by experiment. A protocol is also designed based on the BMH algorithm which is more efficient and conceals more private information.

Secure Multi-party Computation (SMC) has been a research focus in international cryptography community in recent years. SMC deals with the following situation: Two (or many) parties want to jointly perform a computation without disclosing their private inputs. Privacy-preserving convex hulls problem is a special case of SMC and it can be applied in many fields such as military and commercial fields. In this paper, we first present two privacy-preserving protocols to solve the convex hulls problem by using Yao 's millionaire protocol. We also discuss the security, correctness and performance of the two protocols. Based on the Euclid-distance Measure Protocol, an approximate solution to the convex hulls problem is proposed for fairness, which conceals more private information.

When data mining occurs on distributed data, privacy of parties becomes great concerns. This paper considers the problem of mining quantitative association rules without revealing the private information of parties who compute jointly and share distributed data. The issue is an area of privacy preserving data mining (PPDM) research. Some researchers have considered the case of mining Boolean association rules; however, this method cannot be easily applied to quantitative rules mining. A new secure set union algorithm is proposed in this paper, which unifies the input sets of parties without revealing any element's owner and has lower time cost than existing algorithms. The new algorithm takes the advantages of both in privacy-preserving Boolean association rules mining and in privacy-preserving quantitative association mining. This paper also presents an algorithm for privacy-preserving quantitative association rules mining over horizontally portioned data, based on CF tree and secure sum algorithm. Besides, the analysis of the correctness, the security and the complexity of our algorithms are provided

In multiple-precision algorithms, the design and implementation of division is the most complicated. On the basis of some classical algorithms, this paper introduces an efficient improved algorithm. This algorithm omits the most majority of normalization of classical algorithms and uses integer arithmetic instead of floating-point data. By analyzing the algorithm and comparing the arithmetic cost, we conclude that this algorithm is at least three times faster than the most efficient previous solution. Key words: multiple-precision, algorithm, division.

For the problem of decision tree classification with privacy concerns, we propose several efficient secure multi-party computation protocols to construct a privacy preserving ID3 algorithm over horizontally partitioned data among multiple parties. Our algorithm presents the first solution to privacy preserving decision tree classification among more than two parties. We also make a performance comparison with the existing solution, which is only applicable to the twoparty case. The result shows that our solution has a significantly better performance.