With rapid urbanization in the modern era, traffic signals from various sensors have been playing a significant role in monitoring the states of cities, which provides a strong foundation in ensuring safe travel, reducing traffic congestion and optimizing urban mobility. Most existing methods for traffic time series modeling often rely on the original data modality, i.e., numerical direct readings from the sensors in cities. However, this unimodal approach overlooks the semantic information existing in multimodal heterogeneous urban data in different perspectives, which hinders a comprehensive understanding of traffic signals and limits the accurate prediction of complex traffic dynamics. To address this problem, we propose a novel Multimodal framework, MTP, for urban Traffic Profiling, which learns multimodal features through numeric, visual, and textual perspectives in the frequency domain. The three branches drive a multimodal perspective of traffic signal learning for augmentation, while the frequency learning strategies delicately refine the information for extraction. Specifically, we first conduct the visual augmentation for the traffic time series, which transforms the original modality into periodicity images and frequency images for visual learning. Also, we augment descriptive texts for the traffic time series based on the specific topic, background information and item description for textual learning. To complement the numeric information, we utilize frequency multilayer perceptrons for learning on the original modality. We design a hierarchical contrastive learning on the three branches to fuse the three modalities. Finally, extensive experiments on six real-world datasets demonstrate superior performance compared with the state-of-the-art approaches.

Accurate conversion rate (CVR) prediction is critical for recommender systems to capture user conversion intent and increase platform revenues. Traditional CVR models commonly suffer from sample selection bias (SSB) and data sparsity (DS), which has led to the adoption of click-through & conversion rate (CTCVR) multi-task learning frameworks to alleviate these issues. However, existing methods implicitly mislabel some unclicked samples with genuine conversion potential as negatives, thereby exacerbating the false negative sample (FNS) problem. To address this, we propose IdeFN, a multi‑task CVR framework that identifies false negatives in the unclicked space to enable CVR prediction across the entire exposure space and leverages CTR as an auxiliary task for shared‑parameter learning. Specifically, IdeFN consists of two main components, i.e., relaxed partial optimal transport (RPOT) module and sample relabeling mechanism (SRM). The former estimates the soft matching strengths between unclicked samples and positive samples under a relaxed partial optimal transport formulation, establishing corresponding relationships between these samples. The latter adaptively re-labels the unclicked samples according to the derived matching strengths, without relying on static or heuristic thresholds, thus enhancing the reliability of the generated pseudo-labels. Experimental results demonstrate that IdeFN effectively mitigates the FNS problem, achieving substantial improvements in CVR prediction accuracy.

Sequential recommendation has achieved remarkable success across various application domains due to its ability to capture dynamic user preferences. Therefore, its effectiveness significantly diminishes in user cold-start scenarios, where new users have limited or no interaction history. Current solutions typically design specialized model architectures to infer cold-start user preferences from auxiliary information, such as user attributes or social networks. However, such methods overlook compatibility with advanced sequential recommender models, preventing the efficient extraction of sequential features. To address this limitation, we propose CDMRec, a Causal Diffusion Preference Model for user cold-start sequential recommendation. CDMRec generates diffusion-based preference representations for cold-start users, which can be directly utilized by existing sequential recommendation models. The framework first constructs a Preference-Dominant Sequence (PDS) by isolating interactions most indicative of user interests, mitigating noise from irrelevant behaviors. Then, leveraging causal inference, CDMRec identifies key causal variables from PDS to condition the diffusion process, enabling the generation of personalized behavioral preferences. Extensive experiments on three public datasets demonstrate that CDMRec can be seamlessly integrated into mainstream sequential recommender models, yielding substantial performance gains in cold-start settings.

Sequential recommendation aims to predict the next user interaction by modeling historical behavior sequences. Recently diffusion models (DMs) have emerged as a promising generative approach due to their robustness and capacity for uncertainty modeling. However, existing diffusion-based recommendation approaches still encounter two major challenges: sample drift during the noise injection process, which compromises the stability of generation; and limited adaptability to noisy data, which hampers the effectiveness of personalized recommendations. To address these issues, we propose GSDiffRec, a novel generative sequential recommendation approach that integrates two core modules: (i) Semantic-Targeted Guidance Module (STG) built upon an enhanced Transformer backbone equipped with shaped attention and convolutional components to improve representational efficiency and modeling capacity; and (ii) Geodesic Diffusion Module (GDM) enforcing manifold constraints through geodesic random walks, thereby preserving geometric consistency and enhancing denoising stability throughout the diffusion process. Extensive experiments on two public Amazon datasets demonstrate that GSDiffRec significantly outperforms a wide range of competitive baselines. Further ablation studies validate the complementary contributions and effectiveness of the GDM and STG modules.

Abnormal user detection has been a critical and widely studied research problem in social networks since these users can create significant risks to platform security and privacy leakage. Currently, graph-based models are commonly used for exploring the structured social network data and temporally dynamic user interactions, leading to significant advances in dynamic heterogeneous graph-based abnormal user detection. However, most existing approaches are correlation-driven and lack the ability to separate stable patterns from transient noise. Furthermore, these methods are highly dependent on inherent labels and fail to detect common few-shot anomalies in social networks. To address these issues, we propose CIFAD, a Causal-Invariant Few-shot Anomaly Detection method that improves few-shot anomaly detection with an active annotation strategy. Specifically, CIFAD first integrates a sparse lagged attention encoder to model multi-relational temporal interactions. Furthermore, it introduces causal-invariant subspace decomposition to disentangle stable causal signals from dynamic environmental noise and improve generalization. Finally, it designs an active annotation strategy based on influence functions and coverage optimization to maximize the utility of limited labels in a closed-loop process. Extensive experiments on multiple real-world social network datasets demonstrate that our method achieves higher accuracy than state-of-the-art methods, validating its robustness in abnormal user detection for social networks.

With the rapid growth of multi-modal content on the Web, robust vision-language models are essential for semantic understanding and classification of web images under diverse and dynamic contexts, supporting Web applications such as multimedia search and recommendation. Prompt learning has proven effective for enhancing vision-language models in semantic image classification tasks. However, previous methods often suffer from poor generalization: the learned prompts tend to overfit the base classes seen during training, leading to poor performance on unseen classes and under distribution shifts. This issue is especially challenging in Web-scale data, where new classes emerge and distributions shift dynamically. To address these limitations, we propose PLIKD, a novel prompt learning method that integrates instance-aware knowledge distillation for robust Web-scale semantic image classification. Specifically, PLIKD introduces an instance-aware knowledge extraction module, which leverages multi-modal large language models through a step-by-step strategy to extract external knowledge for each image instance. To incorporate this extracted knowledge, PLIKD further introduces an instance-aware knowledge distillation module, which consists of two key steps: (1) a dual-teacher strategy for robust and informative knowledge distillation, and (2) fine-grained cross-modal alignment via Smooth and Sparse Optimal Transport. Extensive experiments demonstrate that PLIKD significantly improves generalization to both seen and unseen classes, and remains robust under distribution shifts, outperforming existing state-of-the-art methods on Web-scale semantic image classification.

Multi-modal recommendation has become essential with the rapid expansion of online platforms such as e-commerce and video-sharing applications. In this work, we focus on the Multi-Modal New Item Recommendation (MMNIR) problem, where items with multi-modal content but newly introduced items lack interaction history. The MMNIR problem is particularly challenging in two aspects: (1) a large number of new items are created rapidly over time without any interaction data, (2) not all existing interactions are equally useful, and it is non-trivial to identify informative behaviors from users with similar preferences. However, previous methods fail to identify users with similar preferences and to capture informative behaviors from historical data. Furthermore, conventional models primarily rely on simple co-occurring signals, leading to spurious neighbors and neglecting the informative behaviors of truly similar users with consistent preferences. To fill this gap, we propose Joint Similar User Exploration and Informative Behavior Guidance (SuperG) for solving the MMNIR problem. SuperG first proposes a similar user exploration module to identify users with similar preferences to the target user. Then it incorporates an informative behavior mining module to retrieve informative behaviors from both the target user and similar users' histories to support new item recommendation. Finally, SuperG proposes a behavior-guided hybrid recommendation module to incorporate the retrieved behavioral signals to guide the recommendation of new items. Our empirical study on three real datasets demonstrates that SuperG outperforms the state-of-the-art models under the MMNIR setting.

Web-scale platforms and online services rely on log-based anomaly detection to safeguard availability, latency SLOs, and user experience. In real-world web interactions, system logs often exhibit irregular temporal intervals, bursty densities, and heterogeneous semantics, which pose significant challenges for log anomaly detection. Existing methods such as LSTM and Transformer assume a fixed input window, which conflicts with the inherently irregular nature of system logs. Moreover, most prior works build a single-view representation, overlooking the multi-relational nature of logs. To overcome these challenges, we propose DyLogNet, a dynamic multi-relational graph framework for log anomaly detection. Specifically, this framework constructs a density-aware dynamic graph with variable-length windows, and represents logs from three relational perspectives: temporal co-occurrence, semantic similarity, and anomaly tendency. Next, we design a cross-layer attention mechanism that integrates heterogeneous structures to highlight the most relevant relations and enhance event representations. Furthermore, a cross-snapshot memory injection module updates global memory through a recurrent unit and injects it into current graph representations via an affine transformation, enabling temporal continuity. Experiments on three public log datasets demonstrate that DyLogNet outperforms state-of-the-art methods, especially in few-shot scenarios.

Personalized federated learning is a practical solution to provide personalized services for Internet of Things devices while protecting their privacy. However, recent research found that personalized federated learning is still vulnerable to attacks, and its privacy can still be compromised during the training process. With an increasing number of devices and more diverse data, information traceability is also much more complicated. To address these problems, this paper proposes pFedBlock, a blockchain-based split federated learning framework to alleviate privacy risks and enhance the traceability during model training. Through the application of blockchain decentralized and immutable characteristics, pFedBlock can save each model update in a secure way and preserve a trustworthy log for all training behaviors. Such design can be beneficial to protect the training process and minimize the possible attacks from adversarial behaviors. Meanwhile, we also design a hybrid aggregation strategy in the federated framework so that devices can perform model updates in a more secure way. Experimental analysis shows that compared with traditional personalized federated learning methods, pFedBlock can obtain better performance in both models performance and system security.

API recommendation is a crucial task in code intelligence, aiming to suggest suitable APIs for programming queries. Recent efforts have integrated Large Language Models (LLMs) into this task. However, these methods overlook the compatibility between recommended APIs and fail to fully utilize the factual knowledge of APIs. Moreover, these prompting-only methods are limited by the insufficient domain-specific knowledge of LLMs. In this article, we propose a novel fine-tuning method, KDRAR, designed to leverage knowledge-driven reasoning with LLMs for compatible and interpretable API recommendation. To fully utilize the factual knowledge, we introduce a dual matching strategy that leverages both function descriptions and keyword matching to retrieve candidate APIs. To handle compatibility, we translate compatibility information into descriptive knowledge, which is integrated into the recommendation process. Furthermore, we adopt a distilled fine-tuning strategy: a student LLM is trained via distillation from a teacher LLM to perform step-by-step reasoning for enhanced recommendation and explanation. By considering both function matching and compatibility information, the knowledge-driven reasoning not only improves API recommendation accuracy but also provides reasonable explanations for the recommendations. Experimental results show that our method significantly outperforms baseline methods on API recommendation tasks across multiple API domains.

This is an erratum for the article “Knowledge-Driven Reasoning for Compatible and Interpretable API Recommendation via Teacher LLM Distillation” published in ACM Trans. Inf. Syst. 44, 1, Article 27 (December 2025), 30 pages.

Traditional methods for ensuring security and privacy face challenges in safeguarding multimedia data within the IoT-edge continuum, as their significant computational demands render them unsuitable for IoT devices with limited resources. Next, we find that the federated learning techique can naturally adapt to edge frameworks and provide effective data security and privacy protection. In this article, we propose FLiForest, an innovative anomaly detection approach that integrates federated learning with the isolation forest algorithm, tailored for the IoT-edge continuum. Specifically, our method designs a three-stage process, including data collection and sampling, model training, and data testing, to joint-train an isolation forest among clients and edge servers. FLiForest facilitates decentralized model training across IoT devices, enhancing data privacy and reducing computational burden, without necessitating the exchange of multimedia data. Through extensive experiments on a variety of multimedia datasets, the efficacy of our method is benchmarked against the state-of-the-art anomaly detection methods, showcasing its superior detection accuracy and robustness in ensuring data privacy and security.

Graph Neural Networks (GNNs) are widely applied on graph-level tasks, such as node classification, link prediction and graph generation. Existing GNNs mostly adopt a message-passing mechanism to aggregate node information with their neighbors, which often makes node information similar after rounds of aggregations and leads to oversmoothing. Although recent works have made improvements by combining different message aggregation methods or introducing semantic encodings as priors, these message-passing based GNNs still fail to combat oversmoothing after multiple iterations of node aggregation. Besides, the feature extraction ability of these methods is restricted because of the graph sparsity that hinders the aggregation of node information. To deal with the above two issues, we propose Neighborhood-based and Label-enhanced Graph Transformer (NLGT), a novel and effective framework for graph learning. Specifically, we present a label-enhanced feature fusion mechanism that integrate the shallow node features and label embeddings as enhanced features. Moreover, we design a neighborhood-based mask attention mechanism to alleviate the negative effects caused by the sparsity of the graph. In the predicting stage, we aggregate the prediction results from multiple sampled sub-graphs and apply voting mechanisms to enhance the accuracy and robustness of our framework. Finally, extensive experiments are conducted on four open benchmark datasets, which demonstrate the effectiveness and robustness of our proposed framework compared with existing state-of-the-art methods.

Graph Contrastive Learning (GCL), as a primary paradigm of graph self-supervised learning, spurs a fruitful line of research in tackling the data sparsity issue by maximizing the consistency of user/item embeddings between different augmented views with random perturbations. However, diversity, as a crucial metric for recommendation performance and user satisfaction, has received rather little attention. In fact, there exists a challenging dilemma in balancing accuracy and diversity. To address these issues, we propose a new Graph Contrastive Learning (DivGCL) model for diversifying recommendations. Inspired by the excellence of the determinant point process (DPP), DivGCL adopts a DPP likelihood-based loss function to achieve an ideal trade-off between diversity and accuracy, optimizing it jointly with the advanced Gaussian noise-augmented GCL objective. Extensive experiments on four popular datasets demonstrate that DivGCL surpasses existing approaches in balancing accuracy and diversity, with an improvement of 23.47% at T@20 (abbreviation for trade-off metric) on ML-1M.

Personalized federated learning (PFL) has recently gained significant attention for its capability to address the poor convergence performance on highly heterogeneous data and the lack of personalized solutions of traditional federated learning (FL). Existing mainstream approaches either perform personalized aggregation based on a specific model architecture to leverage global knowledge or achieve personalization by exploiting client similarities. However, the former overlooks the discrepancies in client data distributions by indiscriminately aggregating all clients, while the latter lacks fine-grained collaboration of classifiers relevant to local tasks. In view of this challenge, we propose a Personalized Federated learning method for Enhancing Collaboration among Similar Classifiers (PFedCS), which aims at improving the client’s accuracy on local tasks. Concretely, it is achieved by leveraging awareness of the client classifier similarities to address the above problems. By iteratively measuring the distance of the classifier parameters between clients and clustering with each client as a cluster center, the central server adaptively identifies the collaborating clients with similar data distributions. In addition, a distance-constrained aggregation method is designed to generate customized collaborative classifiers to guide local training. As a result, extensive experimental evaluations conducted on three datasets demonstrate that our method achieves state-of-the-art performance.

Time series forecasting has significant applications across various domains, including industry, agriculture, and finance. Transformer-based models have shown significant promise in enhancing time series forecasting over the past few years. However, existing methods struggle to simultaneously capture local details and global semantics under single-view architectures. They also find it difficult to dynamically adapt to time-varying and multi-scale temporal patterns while accurately modeling the complex, time-varying relationships between multiple variables. To address these challenges, we propose HRCformer, a novel Transformer-based framework that introduces two key innovations: the Hierarchical Recursive Interaction Convolution (HRIC) and the Triad Adaptive Recalibration Module (TARM). HRIC achieves joint modeling of fine-grained short-term fluctuations and high-order cross-period dependencies in time series by integrating Divide-and-Process Convolution for local processing with Recursive Channel Interaction Convolution for global processing. TARM further enhances dynamic modeling via Dynamic Variance Attention, which amplifies critical temporal deviations through 3D attention, and the Adaptive Multivariate Recalibration, which uses a two-layer fully connected network with nonlinear activation to learn the dynamic relationships between channels, suppresses noise, and emphasizes informative multivariate interactions. Comprehensive experiments conducted on seven real-world datasets highlight the superiority of HRCformer compared to prior state-of-the-art methods.

Document-level relation extraction (RE) aims to extract comprehensive correlations between entities and relations from documents. Most of existing works conduct transfer learning on pre-trained language models (PLMs), which allows for richer contextual representation to improve the performance. However, such PLMs-based methods suffer from incorporating structural knowledge, such as entity-entity interactions. Moreover, current works struggle to infer the implicit relations between entities across different sentences, which results in poor prediction. To deal with the above issues, we propose a novel and effective framework, named DocKS-RAG, which introduces extra structural knowledge and semantic information to further enhance the performance of document-level RE. Specifically, we construct a Document-level Knowledge Graph from the observable documentation data to better capture the structural information between entities and relations. Then, a Sentence-level Semantic Retrieval-Augmented Generation mechanism is designed to consider the similarity in different sentences by retrieving the relevant contextual semantic information. Furthermore, we present a hybrid-prompt tuning method on large language models (LLMs) for specific document-level RE tasks. Finally, extensive experiments conducted on two benchmark datasets demonstrate that our proposed framework enhances all the metrics compared with state-of-the-art methods.

Centralized storage in web-based services poses privacy risks for diverse user data, whereas personalized federated learning (PFL) arnesses the inherent privacy-preserving benefits of FL through distributed collaborative training, avoiding the centralization of sensitive data. Besides, PFL also customizes models to individual client needs, enabling personalized web services on non-independent and identically distributed data across varied environments. Existing PFL techniques often rely on prior knowledge to identify personalization layers, overlooking variable layer sensitivities to heterogeneous data, thus constraining the model's adaptability to diverse statistical distributions and limiting its service effectiveness. To overcome this issue, we introduce FedLD, a layer-wise decoupling method for PFL that precisely captures layer-specific sensitivities to heterogeneous data, enhancing service-oriented model personalization. Firstly, FedLD employs a hypernetwork to evaluate each layer's contribution to performance across varied data distributions, determining the optimal proportion of personalized channels per layer for each client, thus enabling fine-grained, channel-level parameter adjustments tailored to individual service needs. Secondly, knowledge distillation is introduced to enforce consistency between the representations of personalized and shared weights, promoting collaboration between them. Thirdly, an improved aggregation strategy is proposed to enhance collaboration among clients with similar data distributions. Extensive experimental results show that FedLD achieves more effective and generalized results compared to baseline methods across various types of heterogeneous data settings.

Federated Learning (FL) addresses the growing concerns of Internet of Things (IoT) service security and privacy in edge computing environments by enabling collaborative model training without the need to centralize sensitive data. Most existing FL frameworks remain vulnerable to sophisticated threats such as source inference attacks (SIAs), which exploit model updates to infer sensitive information about participating clients, thereby compromising the integrity and security of edge services. To mitigate such attacks and ensure service security and user privacy, various defensive methods, such as RM Learning and RelaxLoss, have been proposed. However, these methods fail to provide effective privacy protection in practical FL scenarios characterized by non-IID data distributions. To address this issue, we propose FedMLU, a novel algorithm designed to counter SIAs effectively. Specifically, FedMLU combines a model alternating update strategy with the RelaxLoss algorithm to minimize the loss discrepancy among samples, thereby reducing the distinguishability exploited by SIAs. Furthermore, distinct soft labels are assigned for training each federated participant model, aiming to decrease the model's prediction confidence and enhance privacy protection. Extensive experiments on synthetic datasets and various real-world datasets demonstrate that our method achieves better defense performance and a more favorable tradeoff between privacy protection and model utility compared to the state-of-the-art RelaxLoss and two popular FL frameworks, particularly in scenarios with data heterogeneity.

Backdoor attacks in vertical federated learning (VFL) are particularly concerning as they can covertly compromise VFL decision-making, posing a severe threat to critical applications of VFL. Existing defense mechanisms typically involve either label obfuscation during training or model pruning during inference. However, the inherent limitations on the defender's access to the global model and complete training data in VFL environments fundamentally constrain the effectiveness of these conventional methods. To address these limitations, we propose the Universal Backdoor Defense (UBD) framework. UBD leverages Label Consistent Clustering (LCC) to synthesize plausible latent triggers associated with the backdoor class. This synthesized information is then utilized for mitigating backdoor threats through Linear Probing (LP), guided by a constraint on Batch Normalization (BN) statistics. Positioned within a unified VFL backdoor defense paradigm, UBD offers a generalized framework for both detection and mitigation that critically does not necessitate access to the entire model or dataset. Extensive experiments across multiple datasets rigorously demonstrate the efficacy of the UBD framework, achieving state-of-the-art performance against diverse backdoor attack types in VFL, including both dirty-label and clean-label variants.

Sequential recommendation generates embedding representations from historical user-item interactions to recommend the next potential interaction item. Due to the complexity and variability of historical user-item interactions, extracting effective user features is quite challenging. Recent studies have employed sequential networks such as time series networks and Transformers to capture the intricate dependencies and temporal patterns in historical user-item interactions, extracting more effective user features. However, limited by the scarcity and suboptimal quality of data, these methods struggle to capture subtle differences in user sequences, which results in diminished recommendation accuracy. To address the above issue, we propose a contrastive learning framework with LLMs-based view augmentation (CLLMRec), which effectively mines differences in behavioral sequences through sample generation. Specifically, CLLMRec utilizes LLMs (Large Language Models) to augment views and expand user behavior sequence representations, providing high-quality positive and negative samples. Subsequently, CLLMRec employs the augmented views for effective contrastive learning, capturing subtle differences in behavioral sequences to suppress interference from irrelevant noise. Experimental results on three public datasets demonstrate that the proposed method outperforms state-of-the-art baseline models, and significantly enhances recommendation performance.

With the rapid urbanization in the modern era, smart traffic profiling based on multimodal sources of data has been playing a significant role in ensuring safe travel, reducing traffic congestion and optimizing urban mobility. Most existing methods for traffic profiling on the road level usually utilize single-modality data, i.e., they mainly focus on image processing with deep vision models or auxiliary analysis on the textual data. However, the joint modeling and multimodal fusion of the textual and visual modalities have been rarely studied in road traffic profiling, which largely hinders the accurate prediction or classification of traffic conditions. To address this issue, we propose a novel multimodal learning and fusion framework for road traffic profiling, named TraffiCFUS. Specifically, given the traffic images, our TraffiCFUS framework first introduces Vision Language Models (VLMs) to generate text and then creates tailored prompt instructions for refining this text according to the specific scene requirements of road traffic profiling. Next, we apply the discrete Fourier transform to convert multimodal data from the spatial domain to the frequency domain and perform a cross-modal spectrum transform to filter out irrelevant information for traffic profiling. Furthermore, the processed spatial multimodal data is combined to generate fusion loss and interaction loss with contrastive learning. Finally, extensive experiments on four real-world datasets illustrate superior performance compared with the state-of-the-art approaches.

Graph anomaly detection (GAD), with its ability to accurately identify anomalous patterns in graph data, plays a vital role in areas such as network security, social media platforms, and fraud detection. Graph autoencoder-based methods are widely used for GAD due to their efficiency and effectiveness in capturing complex patterns and learning meaningful representations. However, the above methods are constrained by hardware memory, hindering the detection for large-scale graph data. In this paper, we propose a Memory-Efficient framework for large-scale attributed Graph Anomaly Detection (MEGAD). Specifically, MEGAD first generates node embeddings and then refines them through a lightweight joint optimization model, ensuring minimal memory overhead. The optimized embeddings are subsequently fed into a detector to compute anomaly scores. Extensive experiments demonstrate that our framework achieves comparable accuracy to state-of-the-art methods across multiple datasets while significantly reducing memory consumption on large-scale graphs.

Federated learning (FL) is a distributed machine learning paradigm that enables multiple clients to collaboratively train a shared model without exposing their raw data. However, existing FL research has primarily focused on optimizing learning performance based on the assumption of uniform client participation, with few studies delving into performance fairness under inconsistent client participation, particularly in model-heterogeneous FL environments. In view of this challenge, we propose PHP-FL, a novel model-heterogeneous FL method that explicitly addresses scenarios with varying client participation probabilities to enhance both model accuracy and performance fairness. Specifically, we introduce a Dual-End Aligned ensemble Learning (DEAL) module, where small auxiliary models on clients are used for dual-end knowledge alignment and local ensemble learning, effectively tackling model heterogeneity without a public dataset. Furthermore, to mitigate update conflicts caused by inconsistent participation probabilities, we propose an Importance-driven Selective Parameter Update (ISPU) module, which accurately updates critical local parameters based on training progress. Finally, we implement PHP-FL on a lightweight FL platform with heterogeneous clients across three different client participation patterns. Extensive experiments under heterogeneous settings and diverse client participation patterns demonstrate that PHP-FL achieves state-of-the-art performance in both accuracy and fairness. Our code is available at: https://github.com/Siyuan01/PHP-FL-main.

The long-tail problem in sequential recommender systems stems from imbalanced interaction data, resulting in suboptimal model performance for tail users and items. Recent studies have leveraged head data to enhance tail data for diminish the impact of the long-tail problem. However, these methods often adopt ad-hoc strategies to distinguish between head and tail data, which fails to capture the underlying distributional characteristics and structural properties of each category. Moreover, due to a substantial representational gap exists between head and tail data, head-to-tail enhancement strategies are susceptible to negative transfer, often leading to a decline in overall model performance. To address these issues, we propose a hierarchical partitioning and stepwise enhancement framework, called HPSERec, for long-tailed sequential recommendation. HPSERec partitions the item set into subsets based on a data imbalance metric, assigning an expert network to each subset to capture user-specific local features. Subsequently, we apply knowledge distillation to progressively improve long-tail interest representation, followed by a Sinkhorn optimal transport-based feedback module, which aligns user representations across expert levels through a globally optimal and softly matched mapping. Extensive experiments on three real-world datasets demonstrate that HPSERec consistently outperforms all baseline methods. The implementation code is available at https://anonymous.4open.science/r/HPSERec-2404.

Code summarization aims to generate concise natural language descriptions of source code, helping developers to acquaint with software systems and reduce maintenance costs. Existing code summarization approaches widely employ attention mechanisms to assess the relevance between nodes in the Abstract Syntax Tree (AST), which generates context vectors that reflect the semantics of the source code. However, these approaches solely relying on AST lack the extraction of features at other levels of granularity, such as code tokens and Control Flow Graph (CFG), which suffer from severe semantic gaps when capturing data and control dependencies. To address this issue, we design an enhanced semantic extractor with multi-granularity feature fusion (MGF-ESE) to improve the model capability in comprehending and processing the overall semantics of the code. Specifically, we present a novel AST generation method that, based on controlling the scale of nodes, introduces syntactic description nodes to raise the semantic density of AST feature. Then we perform both local and global encoding of CFG after embedding the statement nodes. Moreover, through a cross-attention mechanism, we fuse code tokens and CFG with AST to enhance the model's capacity to capture both syntactic and structural information from source code. Finally, extensive experiments on two open-source datasets show that MGF-ESE outperforms the state-of-the-arts with higher-quality code summaries on key metrics, including BLEU, METEOR, and ROUGE-L.

Medical time series has been playing a vital role in real-world healthcare systems as valuable information in monitoring health conditions of patients. Traditional methods towards medical time series classification rely on handcrafted feature extraction and statistical methods; with the recent advancement of artificial intelligence, the machine learning and deep learning methods have become more popular. However, existing methods often fail to fully model the complex spatial dynamics under different scales, which ignore the dynamic multi-resolution spatial and temporal joint inter-dependencies. Moreover, they are less likely to consider the special baseline wander problem as well as the multi-view characteristics of medical time series, which largely hinders their prediction performance. To address these limitations, we propose a Multi-resolution Spatiotemporal Graph Learning framework, MedGNN, for medical time series classification. Specifically, we first propose to construct multi-resolution adaptive graph structures to learn dynamic multi-scale embeddings. Then, to address the baseline wander problem, we propose Difference Attention Networks to operate self-attention mechanisms on the finite difference for temporal modeling. Moreover, to learn the multi-view characteristics, we utilize the Frequency Convolution Networks to capture complementary information of medical time series from the frequency domain. In addition, we introduce the Multi-resolution Graph Transformer architecture to model the dynamic dependencies and fuse the information from different resolutions. Finally, we have conducted extensive experiments on multiple medical real-world datasets that demonstrate the superior performance of our method. Our Code is available at this repository: https://github.com/aikunyi/MedGNN.

In edge computing, the Zero-Trust Security Model (ZTSM), as a key enabling technology for next-generation networks, plays a crucial role in providing authentication for addressing data sharing concerns, such as frequent data breaches, data misuse, and cyberattacks. However, due to the complexity and diversity of edge environments, ZTSM struggles to meet the security requirements of data sharing frameworks solely through enhanced authentication. Consequently, such frameworks with ZTSM still face challenges in ensuring data integrity, evaluating various node behaviors, and coping with the increasing complexity of node attributes. To address these issues, we propose a blockchain-enabled secure, fair and scalable data sharing framework in a zero-trust edge-end environment in this paper. Specifically, we first propose a Merkle forest-based data storage model for the classified storage of loosely coupled data, consequently enhancing the scalability of the model. Then, we design a node behavior-based reputation assessment mechanism to ensure fairness during data sharing. Moreover, a data sharing protocol supervised by smart contract is proposed, working with the aforementioned storage and assessment schemes, to ensure the security of data sharing. Finally, comprehensive security analysis validates the security, fairness and scalability of the proposed framework. Extensive experimental results show that, as transaction volume grows, the time cost of data traversal in the storage model becomes progressively more efficient. Additionally, when the size of the smart contract is increased tenfold, the maximum time cost of the data sharing protocol rises by only 4.98 times.

Embedding-based recommender systems rely on historical interactions to model users, which poses challenges for recommending to new users, known as the user cold-start problem. Some approaches incorporate social networks to deduce preferences based on the social circles of cold-start users to solve the problem of sparse features. However, such methods have difficulty distinguishing between superficial correlations and causal relationships in social behaviors, leading to inaccuracies in predicting user preferences. To address the aforementioned issues, we propose the Causal Contrastive Learning Recommendation (C2lRec) framework. Specifically, we causally model the inference of hidden preferences from the feature and historical behavior of warm users and predict user interactions based on such preferences. The counterfactual inference is subsequently performed to intervene and extract interactions from historical behaviors of warm users that influence their preferences, designating as primary causal variables. Additionally, we utilize the primary causal variables from users within the social circle of cold-start users to substitute the missing historical interactions of cold-start users and employ a similar causal modeling approach to uncover hidden preferences as we do with warm users. Finally, we realize causal contrastive learning to enhance the distribution of cold-start users. Extensive experiments conducted on three public datasets demonstrate that the recommendation performance of C2lRec exceeds that of state-of-the-art methods.

Anomaly detection is a popular research topic in Artificial Intelligence and has been widely applied in network security, financial fraud detection, and industrial equipment failure detection. Isolation forest based methods are the base algorithms to detect anomalies in these scenarios for their simplicity and efficiency, which has been further exploited with multi-folk trees and learning mechanisms to realize the optimal isolation forest for high detection accuracy. However, the optimal isolation forest is time-consuming with the learning mechanisms, resulting in the task failing of time-constrained applications. Moreover, the original optimal isolation forest fails to construct the optimal tree structure restricted by the time complexity. To address the above challenges, we propose an efficient anomaly detection method called EEIF, which realizes the real e-folk structure of the optimal isolation forest in our practical algorithm design. Specifically, we design a distribution that perfectly matches the e-branch theory to construct the optimal isolation forest. Then, we design an FR clustering scheme to achieve fast training of the isolation forest with learning to hash and provide related proofs of accuracy and efficiency. Besides, a parallel algorithm is integrated into our method to reduce prediction time. Finally, extensive experiments are conducted on a large amount of real-world datasets and the results demonstrate that our method significantly improves efficiency while ensuring effectiveness, compared with the state-of-the-art methods.

In modern software systems, log anomaly detection is a key technology to ensure system stability and reliability. The existing log-based anomaly detection methods have make significant development, but they fail to provide high detection accuracy when processing semantic information, especially in terms of semantic noise and the need for task-specific semantic embeddings. To address these issues, we propose a novel anomaly detection method called Contrast-Enhanced Log Anomaly Detection (CELA). The CELA model first enriches the content of log events using large language models and further optimizes the pretrained BERT model through contrastive learning techniques, ensuring that the event representations extracted from logs more accurately reflect potential anomalous patterns. On this basis, we have developed an anomaly detection model based on LSTM, which can effectively learn and identify anomalies from rich event representations. Evaluated on two public log datasets, the CELA model demonstrated superior performance compared to existing methods.

In the question-answering system, users frequently express multiple intents within a single utterance. The majority of intent recognition models tend to either primarily address single-intent scenarios or simply aggregate the overall intent context vectors of all tokens, neglecting the integration of multi-intent information. However, current methods suffer from slow inference speed, limited generalization capability, and the relatively independent treatment of multi-intent recognition tasks, leading to suboptimal model performance. In this paper, we design a novel model based on BERT, employing a non-autoregressive approach for the task of multi-intent recognition. This model achieves enhanced speed and accuracy. Additionally, this model introduces a global slot-intent interaction layer, simulating interactions between multiple intents and slots within an utterance. Experimental results on a series of benchmarking datasets demonstrate that this model outperforms the state-of-the-art methods, improving both the effectiveness and efficiency of the overall system.

Social perception recommendation systems can effectively alleviate the user cold start problem by leveraging the side information of social networks. However, most social perception recommendation systems treat user relations as independently existing entities for learning, thereby overlooking potential connections between relations. Additionally, as the number of relations increases, it inevitably imposes a significant computational burden on the servers. To address these issues, we propose the Social perception recommendation based on relational clustering(SPRRC). SPRRC conducts relational clustering of social networks and projects knowledge graphs, effectively mining information about the similarity between relations. First, we cluster relationships in item knowledge graphs and social networks through unsupervised learning. After that, we use local weighted smoothing to aggregate the map information of the clustered items and social network virtual subgraphs respectively, and use the attention networks to learn the representation of users and items in the interactive bipartite graph. Finally, a large number of experiments have verified the high accuracy of our method, compared with the latest methods.

The rapid advancements in vehicular networking technology have enabled onboard users to access a variety of emerging services like precise navigation and real-time hazard avoidance. However, as the diversity and volume of required data expand and demands intensify, most of vehicular services are unable to satisfy user expectations. Although crowdsensing architectures based on edge computing have been proposed in vehicular networks, how to optimally allocate the sensing service time of roadside units and set appropriate prices for services to maximize user benefits still remain significant challenges. To solve the above issue, in this paper, a crowdsensing service pricing method is proposed in vehicular edge computing. Specifically, a vehicular networking crowdsensing framework based on the edge computing is designed. Then, the optimization problem of perception time pricing during the crowdsensing process is modeled into a Stackelberg game and further the multi-agent deep deterministic policy gradient-based algorithm is employed to obtain the optimal pricing strategy with user profits maximization. Finally, simulation results demonstrate that the proposed method significantly increases the payoff for all participants during the crowdsensing process.

Intelligent connected vehicles (ICV) are a key area of competition that is a new driving force for global technological change and technological innovation in vehicles, and the development of its operating system has become an increasingly important priority. As the transformation of vehicles electronic and electrical architecture accelerates, the cross-domain integration of vehicles operating systems has become a general trend. As the main technology for cross-domain integration, virtualization technology is bound to become a battleground for various vehicle manufacturers. This paper aims to take an in-depth look at virtualization of automotive operating systems. Specifically, the overall architecture of operating system virtualization is described first, and then the key functional modules are analyzed in detail, including CPU, memory, interrupts, I/O devices and GPU virtualization technology. Then, a API forwarding-based UV-GPU virtualization scheme is especially designed whose implement details and development prospect in ICV are deeply discussed.

Anomaly detection is a critical aspect of industrial production processes. Most of the self-supervised training system utilise Convolutional Neural Networks (CNNs) for feature extraction. However, these methods often exhibit poor detection performance on small structural anomalies due to the limited receptive field of CNN. The prompt learning method has been demonstrated to be an effective means of enhancing the global information extraction abilities of the model. However, this approach necessitates the input of a significant number of manual prompts. To address these issues, we propose a Memory And Prompt Learning Based Network (MPLNet) for anomaly detection. MPLNet obtains the differences information between normal and detected images by comparing their features and then uses these differences to generate prompts. By automatically generating prompts, it reduces the workload of manually entering prompts in prompt learning. At the same time, it effectively enhances the model’s ability to extract and use global information. Extensive experiments have shown that the proposed MPLNet achieves state-of-the-art anomaly detection performance on the widely used and challenging MVTec AD dataset and MVTec AD-3D dataset. Index Terms—Anomaly detection, U-Net, Transformer, Prompt, artificial intelligence System

Sequential recommendation models generate embeddings for items through the analysis of historical user-item interactions and utilize the acquired embeddings to predict user preferences. Despite being effective in revealing personalized preferences for users, these models heavily rely on user-item interactions. However, due to the lack of interaction information, new users face challenges when utilizing sequential recommendation models for predictions, which is recognized as the cold-start problem. Recent studies, while addressing this problem within specific structures, often neglect the compatibility with existing sequential recommendation models, making seamless integration into existing models unfeasible.To address this challenge, we propose CMCLRec, a Cross-Modal Contrastive Learning framework for user cold-start RECommendation. This approach aims to solve the user cold-start problem by customizing inputs for cold-start users that align with the requirements of sequential recommendation models in a cross-modal manner. Specifically, CMCLRec adopts cross-modal contrastive learning to construct a mapping from user features to user-item interactions based on warm user data. It then generates a simulated behavior sequence for each cold-start user in turn for recommendation purposes. In this way, CMCLRec is theoretically compatible with any extant sequential recommendation model. Comprehensive experiments conducted on real-world datasets substantiate that, compared with state-of-the-art baseline models, CMCLRec markedly enhances the performance of conventional sequential recommendation models, particularly for cold-start users.

Summary With the development of artificial intelligence, cloud‐edge computing and virtual reality, the industrial design that originally depends on human imagination and computing power can be transitioned to metaverse applications in smart manufacturing, which offloads the services of metaverse to cloud and edge platforms for enhancing quality of service (QoS), considering inadequate computing power of terminal devices like industrial sensors and access points (APs). However, large overhead and privacy exposure occur during data transmission to cloud, while edge computing devices (ECDs) are at risk of overloading with redundant service requests and difficult central control. To address these challenges, this paper proposes a minority game (MG) based cloud‐edge service offloading method named COM for metaverse manufacturing. Technically, MG possesses a distribution mechanism that can minimize reliance on centralized control, and gains its effectiveness in resource allocation. Besides, a dynamic control of cut‐off value is supplemented on the basis of MG for better adaptability to network variations. Then, agents in COM (i.e., APs) leverage reinforcement learning (RL) to work on MG history, offloading decision, QoS mapping to state, action and reward, for further optimizing distributed offloading decision‐making. Finally, COM is evaluated using a variety of real‐world datasets of manufacturing. The results indicate that COM has 5.38% higher QoS and 8.58% higher privacy level comparing to benchmark method.

In the social production system, image data are rapidly generated from almost all fields such as factories, hospitals, and transportation, promoting higher requirements for image anomaly detection technologies, including low consumption, higher adaptability, and accuracy. However, existing anomaly detection methods are fragile to heterogeneous image data generated by complex social production systems and tend to require strong computing power and resource support. To address the above problems, a knowledge-driven anomaly detection framework is proposed, in which a local feature enhancement method is designed to strengthen the knowledge representation of the initial features extracted from images. The attention mechanism in deep learning is introduced to adjust the feature attention dynamically according to prior knowledge, which solves the problem of feature loss in the cascade training. To verify the effectiveness of the proposed framework, extensive experiments on social production datasets are conducted. The results demonstrate that our framework outperforms the selected methods on image datasets with different complexity and sample distributions.

Anomaly detection is one of the crucial research topics in artificial intelligence, encompassing various fields such as health monitoring, network intrusion detection, and fraud detection in financial transactions. Deep anomaly detection (DAD) methods are considered as the effective approaches for addressing complex anomaly detection problems. Among them, the deep isolation forest methods have gained rapid development recently due to their simplicity in parameter turning and efficiency in model training. The existing deep isolation forest approaches are all based on representation learning, while OptiForest theoretically proves the crucial role of the tree structure in isolation forest based methods. In this paper, we analyse the search space of isolation trees under specific data instances and address the challenges in finding optimal isolation forest. Based on the theoretical underpinning and genetic algorithm, we design a deep model DOIForest with two mutation schemes and solution selection, which learns the optimal isolation forest and optimises the parameters in data partitioning. Extensive experiments on both synthetic dataset and a series of real-world datasets demonstrate that our approach can achieve better detection accuracy and robustness than the state-of-the-arts.

Edge computing has emerged as a transformative paradigm with applications ranging from IoT to smart homes and transportation systems. However, its decentralized nature presents significant challenges in securing sensitive data, particularly in scenarios involving data sharing. To address these issues, this paper introduces a secure data access control system for edge computing by leveraging blockchain to foster mutual trust among edge servers and employing attribute-based encryption for secure data sharing. Our proposed solution seeks to enhance data security in distributed environments. We validate our method through practical experiments conducted on multiple edge servers, confirming its effectiveness in safeguarding data integrity and privacy.

With the development of modern technology, the application of various types of devices in life has become more extensive, especially with the emergence of the Internet of Things (IoT), which makes a large number of devices to be connected to the network. However, while bringing convenience to life, attacks against IoT devices have also begun to appear and become one of the most concerned issues. These IoT devices produce and transmit data containing important user information every second. Attackers are targeting this characteristic to initiate further malicious attacks on these valuable data. In order to protect users from such attacks, it is important that threats can be detected and identified in time before they cause damage. To address this issue, this paper first reviewed the current state-of-the-art anomaly detection methods and based on the finding of uncovered areas of existing methods, this paper proposed a new anomaly detection framework leveraging the combination of Federated Learning and Isolation Forest. The framework performs tree construction on the clients’ end and further uploads the encrypted data containing the nodes’ information of the trees to the central server for the forest construction, after multiple interactions, the abnormal behaviour in the clients could be able to be identified more effectively and ultimately improved the accuracy of the detection results while protecting the privacy of the client data.

Anomaly detection plays an increasingly important role in various fields for critical tasks such as intrusion detection in cybersecurity, financial risk detection, and human health monitoring. A variety of anomaly detection methods have been proposed, and a category based on the isolation forest mechanism stands out due to its simplicity, effectiveness, and efficiency, e.g., iForest is often employed as a state-of-the-art detector for real deployment. While the majority of isolation forests use the binary structure, a framework LSHiForest has demonstrated that the multi-fork isolation tree structure can lead to better detection performance. However, there is no theoretical work answering the fundamentally and practically important question on the optimal tree structure for an isolation forest with respect to the branching factor. In this paper, we establish a theory on isolation efficiency to answer the question and determine the optimal branching factor for an isolation tree. Based on the theoretical underpinning, we design a practical optimal isolation forest OptIForest incorporating clustering based learning to hash which enables more information to be learned from data for better isolation quality. The rationale of our approach relies on a better bias-variance trade-off achieved by bias reduction in OptIForest. Extensive experiments on a series of benchmarking datasets for comparative and ablation studies demonstrate that our approach can efficiently and robustly achieve better detection performance in general than the state-of-the-arts including the deep learning based methods.

With the great success of deep neural networks (DNNs) in a variety of fields, deep learning gains a pioneering development in anomaly detection. Although deep learning achieves good accuracy in anomaly detection, it is troubled with long execution time and high memory consumption. These problems are associated with the inherent drawbacks of deep learning, such as too many parameters and deep training layers. To remedy the above drawbacks, we try to explore an unsupervised non-neural network deep model for anomaly detection based on the experience of the deep forest. In this paper, we propose a deep anomaly detection framework with hashing based isolation forest (DeepiForest) to achieve effective and robust anomaly detection. Specifically, DeepiForest utilizes hashing based isolation forest and tree-embedding scheme to provide enhanced features and apply multi-layer cascaded architecture to establish a deep framework. DeepiForest inherits the advantages of deep forests, i.e., the framework holds fewer hyper-parameters and smaller model complexity than DNNs, simultaneously producing robust accuracy on anomaly detection. Extensive experiments on different-scale datasets illustrate the efficiency of DeepiForest and its comparable effectiveness to the state-of-the-art deep anomaly detection (DAD) methods.

Abstract Anomaly detection plays a crucial role in many Internet of Things (IoT) applications such as traffic anomaly detection for smart transportation and medical diagnosis for smart healthcare. With the explosion of IoT data, anomaly detection on data streams raises higher requirements for real-time response and strong robustness on large-scale data arriving at the same time and various application fields. However, existing methods are either slow or application-specific. Inspired by the edge computing and generic anomaly detection technique, we propose an isolation forest based framework with dynamic Insertion and Deletion schemes (IDForest), which can incrementally update the forest to detect anomalies on data streams. Besides, IDForest is deployed on edge servers in parallel through packing each tree into a subtask, which facilitates the fast anomaly detection on data streams. Extensive experiments on both synthetic and real-life datasets demonstrate the efficiency and robustness of our framework for anomaly detection.

Recently, biometric identification has been extensively used for border control. Some face recognition systems have been designed based on Internet of Things. But the rich personal information contained in face images can cause severe privacy breach and abuse issues during the process of identification if a biometric system has compromised by insiders or external security attacks. Encrypting the query face image is the state-of-the-art solution to protect an individual’s privacy but incurs huge computational cost and poses a big challenge on time-critical identification applications. However, due to their high computational complexity, existing methods fail to handle large-scale biometric repositories where a target face is searched. In this article, we propose an efficient privacy-preserving face recognition scheme based on clustering. Concretely, our approach innovatively matches an encrypted face query against clustered faces in the repository to save computational cost while guaranteeing identification accuracy via a novel multi-matching scheme. To the best of our knowledge, our scheme is the first to reduce the computational complexity from O(M) in existing methods to approximate O (√ M ), where M is the size of a face repository. Extensive experiments on real-world datasets have shown the effectiveness and efficiency of our scheme.

Anomaly detection is one of the most important data mining tasks in many real-life applications such as network intrusion detection for cybersecurity and medical diagnosis for healthcare. In the big data era, these applications demand fast and versatile anomaly detection capability to handle various types of increasingly huge-volume data. However, existing detection methods are either slow due to high computational complexity, or unable to deal with complicated anomalies like local anomalies. In this paper, we propose a novel anomaly detection method named OPHiForest with the use of the order preserving hashing based isolation forest. The core idea is to learn the information from data to construct better isolation forest structure than the state-of-the-art methods like iForest and LSHiForest, which can achieve robust detection of various anomaly types. We design a fast two-step learning process for the order preserving hashing scheme. This leads to stronger order preservation for better hashing, and therefore enhances anomaly detection robustness and accuracy. Extensive experiments on both synthetic and real-world data sets demonstrate that our method is highly robust and scalable.

With the advent of IoT (Internet of Things) age, considerable web services are emerging rapidly in service communities, which places a heavy burden on the target users' service selection decisions. In this situation, various techniques, e.g., collaborative filtering (i.e., CF) is introduced in service recommendation to alleviate the service selection burden. However, traditional CF-based service recommendation approaches often assume that the historical user-service quality data is centralized, while neglect the distributed recommendation situation. Generally, distributed service recommendation involves inevitable message communication among different parties and hence, brings challenging efficiency and privacy concerns. In view of this challenge, a novel privacy-preserving distributed service recommendation approach based on Locality-Sensitive Hashing (LSH), i.e., DistSRLSH is put forward in this paper. Through LSH, DistSRLSH can achieve a good tradeoff among service recommendation accuracy, privacy-preservation and efficiency in distributed environment. Finally, through a set of experiments deployed on WS-DREAM dataset, we validate the feasibility of our proposal in handling distributed service recommendation problems.

Mobile cloud computing provides powerful computing and storage capacity on managing GPS big data by offloading vast workloads to remote clouds. For the mobile applications with urgent computing or communication deadline, it is necessary to reduce the workload transmission latency between mobile devices and clouds. This can be technically achieved by expanding mobile cloudlets that are moving co-located with Access Points (APs). However, it is not-trivial to place such movable cloudlets efficiently to enhance the cloud service for dynamic context-aware mobile applications. In view of this challenge, an adaptive cloudlet placement method for mobile applications over GPS big data is proposed in this paper. Specifically, the gathering regions of the mobile devices are identified based on position clustering, and the cloudlet destination locations are confirmed accordingly. Besides, the traces between the origin and destination locations of these mobile cloudlets are also achieved. Finally, the experimental results demonstrate that the proposed method is both effective and efficient.