Multimodal Sentiment Analysis (MSA) aims to infer human sentiment by integrating information from multiple modalities such as text, audio, and video. In real-world scenarios, however, the presence of missing modalities and noisy signals significantly hinders the robustness and accuracy of existing models. While prior works have made progress on these issues, they are typically addressed in isolation, limiting overall effectiveness in practical settings. To jointly mitigate the challenges posed by missing and noisy modalities, we propose a framework called Two-stage Modality Denoising and Complementation (TMDC). TMDC comprises two sequential training stages. In the Intra-Modality Denoising Stage, denoised modality-specific and modality-shared representations are extracted from complete data using dedicated denoising modules, reducing the impact of noise and enhancing representational robustness. In the Inter-Modality Complementation Stage, these representations are leveraged to compensate for missing modalities, thereby enriching the available information and further improving robustness. Extensive evaluations on MOSI, MOSEI, and IEMOCAP demonstrate that TMDC consistently achieves superior performance compared to existing methods, establishing new state-of-the-art results.

Accurate inference of absolute copy numbers beyond simple gains and losses from single-cell chromatin accessibility (scATAC-seq) data remains challenging, thereby obscuring the distinction between genetic and epigenetically driven oncogenic dependencies. Here, we present TeaCNV, a computational framework that reconstructs clonal absolute copy number profiles and tumor clonal architectures from scATAC-seq data without matched DNA baselines. Through validation both in silico and against bulk whole-genome sequencing in renal cell carcinomas, TeaCNV resolved subclonal absolute copy number profiles with less than 10% error and detected copy number variations (CNVs) with 98.6% accuracy, outperforming existing methods. Applied to six cancer types including renal, breast, pancreatic, head and neck, colorectal, and ovarian cancers, TeaCNV delineated polyclonal architectures and revealed distinct chromatin accessibility patterns driven by CNVs in key driver genes, including AKT2, ZNF217, and SOX2. By enabling absolute copy number profiling and clonal deconvolution from epigenomic assays, TeaCNV bridges critical gaps in studying oncogenic dependencies and genotype-phenotype relationships at single-cell resolution.

Underwater communication is essential for environmental monitoring, marine biology research, and underwater exploration. Traditional underwater communication faces limitations like low bandwidth, high latency, and susceptibility to noise, while semantic communication (SC) offers a promising solution by focusing on the exchange of semantics rather than symbols or bits. However, SC encounters challenges in underwater environments, including semantic information mismatch and difficulties in accurately identifying and transmitting critical information that aligns with the diverse requirements of underwater applications. To address these challenges, we propose a novel SC framework based on Large Language Models (LLMs). Our framework leverages visual LLMs to perform semantic compression and prioritization of underwater image data according to the query from users. By identifying and encoding key semantic elements within the images, the system selectively transmits high-priority information while applying higher compression rates to less critical regions. On the receiver side, an LLM-based recovery mechanism, along with Global Vision ControlNet and Key Region ControlNet networks, aids in reconstructing the images, thereby enhancing communication efficiency and robustness. Our framework reduces the overall data size to 0.8% of the original. Experimental results demonstrate that our method significantly outperforms existing approaches, ensuring high-quality, semantically accurate image reconstruction.

Multimodal sentiment analysis (MSA) with missing modalities involves understanding the person's sentiment using multimodal data where some modalities are missing. Most existing methods focus on reconstructing the missing modalities using the available modalities from each sample, relying on modality-common information. However, these methods overlook the modality-specific information that other samples can provide. Additionally, these approaches often require the guidance of full modality representations during the reconstruction process, which is impractical in resource-constrained real-world scenarios. To address these challenges, we propose theIntra-sample andIntra-modalEnhancement (IIE) framework. The IIE framework enhances both sample-level and modality-level representations to capture additional modality-common and modality-specific information from existing modalities, without requiring full modalities. Specifically, IIE first learns sample-level representations by distilling modality-common information from the available modalities into learnable latent units. Then, it enhances modality-level representations by leveraging modality-specific information from other samples with the same modality, which is crucial for improving robustness in the presence of missing modalities. Finally, IIE ensures consistency between the enhanced modality-level and sample-level representations, combining the enhanced and initial representations to make predictions. Extensive experiments on three datasets demonstrate that the IIE framework significantly outperforms existing methods in terms of both effectiveness and robustness in handling MSA with missing modalities. Code is available athttps://github.com/YetZzzzzz/IIE.

With the intensification of energy crises and climate change, intelligent microgrid management has become critical for achieving energy resilience and emission reduction. Traditional deep reinforcement learning (DRL) methods excel in optimizing power dispatch and storage utilization but suffer from insufficient interpretability due to their black-box nature. This paper proposes a hybrid framework integrating large language models (LLMs) with DRL to enhance both decision-making performance and interpretability in microgrid management. Leveraging a multi-agent interactive scenario and a modular reward function, the framework balances supply-demand matching, grid interaction, and energy storage coordination. Through carefully engineered prompts, LLMs refine DRL actions and provide real-time natural language explanations for control adjustments. Experimental results demonstrate that the LLM+DRL composite model improves the overall performance score by 0.5%–1.6% compared to standalone DRL while enabling transparent reasoning. The study also explores LLMs scheduling capabilities under varying prior knowledge, highlighting the importance of contextual information for decision quality. This work offers a novel pathway for intelligent and interpretable microgrid management systems, bridging the gap between data-driven optimization and operational transparency.

Social media popularity prediction is increasingly crucial for optimizing user engagement and guiding content recommendation systems. However, existing methods suffer from an excessive reliance on user information, which disproportionately influences predictions and leads to the neglect of content diversity. This oversight results in user bias, which adversely impacts the accuracy of predictions. In this paper, an approach named Feature Disentangling Dual-Stream Network (FDDN) is introduced to address this gap. In FDDN, we introduce the Multimodal Extraction Module to extract content features from different modalities. Additionally, the User Popularity Extraction Module helps to analyze the impact of user features on popularity. The Disentangled Adaptation Module distinguishes the impact of user features from content features, thus alleviating user bias and ensuring a more comprehensive prediction. Extensive experiments on a large public dataset demonstrate the robustness and effectiveness of our approach, indicating superior performance compared to existing methods.

This paper introduces a novel approach named the Privacy-Preserving Socio-Aware Model (PSocLF) for Short-Term Residential Load Forecasting, which addresses the need for forecasting models tailored to district-specific socio-demographic characteristics. By leveraging sociodemographic characteristics and personalized socio-aware knowledge sharing, PSocLF develops district-level forecasting models that enhance load forecasting precision while safeguarding individual privacy. Within PSocLF, we propose a new model structure named the Self-Gating TSMixer (SGTSMixer), which integrates self-gating mixing procedures with stacked multi-layer perceptrons (MLPs). It can efficiently extract temporal patterns and incorporate socioaware information to improve prediction accuracy. Simulation results based on real-world data demonstrate the effectiveness of the proposed PSocLF framework, outperforming alternative training paradigms and benchmarks in model structure design, particularly in scenarios with varying sociodemographic characteristics among districts. This paper contributes to advancing federated residential load forecasting and highlights the practical benefits of integrating sociodemographic information for improved forecasting accuracy and effectiveness.

Modern electricity theft poses severe risks to power grid stability, particularly as cyber-attacks targeting Advanced Metering Infrastructure (AMI) become increasingly covert. This paper proposes a deep learning framework for Electricity Theft Detection (ETD) based on a Transformer encoder and a Dynamic-Weight Multi-Head Classifier (DW-MHC). The model extracts temporal load features via self-attention and employs specialized heads to capture diverse theft patterns, such as abrupt anomalies and gradual deviations. A soft-attention fusion mechanism adaptively integrates the head outputs for robust prediction. The framework also accommodates variability across residential and industrial users, whose load profiles may differ significantly in scale and regularity. Experimental results demonstrate the model's superior performance in detecting varied theft behaviors among consumers, achieving improved performance over traditional single-head and static classifiers.

As social media becomes a dominant platform for sharing content, predicting the popularity of user posts has become increasingly important for applications such as content recommendation, trend forecasting, and user engagement. However, this task is challenging due to the diverse and multimodal nature of social media posts, which often include unstructured text, images, and structured metadata. To address this challenge, we propose Fusion-Aware Multi-modal Ensemble (FAME), a framework effectively captures and integrates diverse information sources within social media content. Unlike prior approaches that rely on a single model to process all modalities, FAME leverages four specialized predictors. Three of them-CatBoost, LightGBM, and AutoGluon-are tree-based models that excel at handling structured metadata and its interactions with unstructured features. The fourth is a denoising autoencoder (DAE), which learns robust joint representations from unstructured text and image data. These models are combined through a weighted ensemble strategy, allowing FAME to leverage the complementary strengths of different architectures. Experiments on the Social Media Prediction Dataset demonstrate that FAME significantly outperforms existing baselines, achieving state-of-the-art results and validating its effectiveness in modeling the complex, multimodal nature of social media content.

Multimodal Sentiment Analysis (MSA) aims to infer human emotions by integrating complementary signals from diverse modalities. However, in real-world scenarios, missing modalities are common due to data corruption, sensor failure, or privacy concerns, which can significantly degrade model performance. To tackle this challenge, we propose Hyper-Modality Enhancement (HME), a novel framework that avoids explicit modality reconstruction by enriching each observed modality with semantically relevant cues retrieved from other samples. This cross-sample enhancement reduces reliance on fully observed data during training, making the method better suited to scenarios with inherently incomplete inputs. In addition, we introduce an uncertainty-aware fusion mechanism that adaptively balances original and enriched representations to improve robustness. Extensive experiments on three public benchmarks show that HME consistently outperforms state-of-the-art methods under various missing modality conditions, demonstrating its practicality in real-world MSA applications.

Cancer is characterized by aneuploidy, often resulting from chromosomal instability (CIN) due to events such as whole-genome duplication (WGD), breakage-fusion-bridge (BFB) cycles, and chromothripsis. These catastrophic events drive de novo copy number alterations (CNAs) with different chromosomal rearrangement patterns, contributing to tumor heterogeneity and evolution. The lack of methods capable of dissecting complex patterns of CNAs at single-cell resolution limits the understanding of how chromosome aneuploidy influences tumor evolution. To address this, we developed ChromoPattern, a computational framework to quantify four distinct chromosomal rearrangement patterns associated with complex CNA mechanisms based on single-cell copy number profiles. ChromoPattern effectively differentiates these patterns through simulations and real-world data. We applied ChromoPattern to single-cell DNA sequencing data from a liver cancer cell line and breast cancer patients. Our analysis identified the oscillation pattern of CNAs as a significant contributor to tumor heterogeneity and evolution. This pattern was associated with a 20% increase in the de novo CNA rate during tumor evolution and was linked to increased intratumor heterogeneity and subclonality. Further analysis using single-cell RNA sequencing data demonstrated that tumor clones with an enriched oscillation pattern display increased cell cycle activity and a stem-like phenotype. These findings underscore the significant and differential roles of chromosomal rearrangement patterns in tumor heterogeneity and evolution. The identification of the oscillation pattern as a key driver of tumor heterogeneity and evolution opens new avenues for developing targeted therapies and diagnostic tools.

This paper addresses a class of sliding mode control (SMC) strategies for high-frequency sampled systems with uncertainties, utilizing a generalized Delta operator reaching law. First, Delta operator sampled systems are regarded as a bridge in the unified research framework of discrete-time systems and continuous-time systems. With the increase of sampling frequency, Delta operator sampled systems can maintain consistent stability with their corresponding continuous models. Then, a generalized reaching law by Delta operator is presented, which can degenerate into multiple existed reaching laws under certain conditions. And then, the reaching condition, quasi-sliding mode (QSM) band, and finite time reachability of the proposed generalized Delta operator reaching law are analyzed. Next, sliding mode controllers are designed separately for both matched uncertainty and unmatched uncertainty cases. With the effect of these controllers, Delta operator SMC systems exhibit desirable dynamic characteristics. Finally, two examples on truck-trailer system and ball-beam system demonstrate the effectiveness of Delta operator sliding mode controllers, and show efficient performance in stabilizing high-frequency sampled systems with uncertainties. These scenarios illustrate the potential of the proposed strategies for practical applications. Note to Practitioners—This work aims to develop sliding mode control strategy for Delta operator sampled systems, which is of fundamental significance in the fields of high-frequency sampled control. We propose a generalized Delta operator reaching law and provide a unified framework for high-frequency sampled discrete-time systems and their corresponding continuous-time systems, addressing practical challenges in areas, such as high-speed motors, supercomputing, and network transmission. The reaching condition, quasi-sliding mode bandwidth, and finite time reachability of the proposed generalized Delta operator reaching law are fully considered. Given that the effects of unknown matched or unmatched uncertainties, sliding mode controllers based on Delta operator are designed, which are applied to truck-trailer system and ball-beam system.

With billions of users worldwide, accurately predicting social media popularity is crucial for assessing user behavior, forecasting trends, and enhancing social interactions and business strategies. However, this task presents significant challenges. Firstly, the extraction of valuable insights is complicated by the presence of tri-modal data (visual, text, structured) and pervasive noise. Secondly, the applicability of knowledge acquired during the pre-training phase is often limited due to discrepancies with downstream prediction tasks during the fine-tuning phase. Existing methods for Social Media Popularity Prediction (SMPP), including traditional models and Visual-and-language Models (VLMs), struggle to overcome these challenges, thereby failing to achieve satisfactory accuracy. To tackle these challenges, we propose a novel approach named Tri-Modal Transformers with Mixture-of-Modality-Experts (TTME) for SMPP. TTME integrates Artificial Intelligence Generated Content to mitigate data noise and incorporate a mix of Modality Experts in pre-training phases to effectively utilize tri-modal data. Moreover, to address training disparity, we explore strategies for downstream task adaptation including the integration of diverse pre-training experts and the implementation of DistillSoftmax. Through empirical evaluation, we demonstrate that the TTME significantly improves the accuracy of social media popularity predictions, effectively utilizes tri-modal data with noise, and enhances transferring knowledge from pre-training to downstream tasks.

Heterogeneity arising from label distribution skew and data scarcity can cause inaccuracy and unfairness in intelligent communication applications that heavily rely on distributed computing. To deal with it, this article proposes a novel personalized federated learning algorithm, named federated contrastive shareable representations (FedCoSRs), to facilitate knowledge sharing among clients while maintaining data privacy. Specifically, the parameters of local models' shallow layers and typical local representations are both considered as shareable information for the server and are aggregated globally. To address performance degradation caused by label distribution skew among clients, contrastive learning is adopted between local and global representations to enrich local knowledge. Additionally, to ensure fairness for clients with scarce data, FedCoSR introduces adaptive local aggregation to coordinate the global model involvement in each client. Our simulations demonstrate FedCoSR's effectiveness in mitigating label heterogeneity by achieving accuracy and fairness improvements over existing methods on datasets with varying degrees of label heterogeneity.

Multimodal sentiment analysis has garnered increasing attention. The bulk of existing work in multimodal sentiment analysis primarily focuses on designing various networks to align and subsequently fuse representations from individual modalities. Contrastive learning, recognized for its intrinsic alignment capabilities, has also been extensively applied in multimodal sentiment analysis. However, current contrastive learning methods are often limited to pairwise modalities and typically perform contrastive learning prior to modality fusion, neglecting the consistency of interactions across multiple modalities. Moreover, they overlook the overall consistency within samples. To address these issues, we introduce a novel Multi-Level Contrastive Learning (MLCL) framework for multimodal sentiment analysis, composed of Uni-Modal Contrastive Learning (UMCL), Bi-Modal Contrastive Learning (BMCL) and Tri-Modal Contrastive Learning (TMCL). UMCL enhances intra-modal representations by creating positive pairs using modality-specific random dropout, while BMCL leverages the asymmetry of attention mechanisms, using two directional attentions as positive samples. TMCL aligns non-overlapping uni-modal and bi-modal representations, underscoring the complementarity of tri-modal information. The effectiveness of MLCL is demonstrated through its performance on multiple datasets. Our comprehensive experiments across multiple datasets demonstrate the superiority of the MLCL framework, which achieves new state-of-the-art performance.

Social media has emerged as a vital platform for communication, information sharing, and acquisition. Predictive analysis of social media data has wide applications, such as sentiment examination and social network analysis. However, existing work often directly utilizes social media data for training, neglecting the issue of mismatched text and images. This neglect can lead to confusion about the contents, thereby affecting the identification of trending topics and the accuracy of social media predictions. In this paper, an approach named Dual-Stream Pre-training Transformer (DSPT) is introduced to address this gap. In DSPT, we use a Visual-Language Model (VLM) and a Language Model (LM) to separately learn from image and text data, mitigating the impact of text-image mismatches. Moreover, to enhance the understanding of the model to social media data, we conduct incremental pre-training for both models. To achieve better feature interaction, we construct an integrated regression module combining LightGBM and CatBoost, jointly predicting the extracted feature embeddings. This dual-stream multimodal feature extraction method improves the performance of predictive tasks. Experimental results validate the effectiveness of our approach, demonstrating its potential and providing deeper insights into multimodal data mining in social media.

Multimodal Sentiment Analysis (MSA) has witnessed remarkable progress and gained increasing attention in recent decade. However, current MSA methodologies primarily rely on global representations extracted from different modalities, such as the mean of all token representations, to construct sophisticated fusion networks. These approaches often overlook the valuable details present in local representations, which consist of fused representations of consecutive several tokens. Additionally, the integration of multiple local representations, and the fusion of local and global information present significant challenges. To address these limitations, we propose the Global-Local Modal (GLoMo) Fusion framework. It comprises two essential components: (i) modality-specific mixture of experts layers that integrate diverse local representations within each modality, and (ii) a global-guided fusion module that effectively combines global and local representations. The former component leverages specialized expert networks to automatically select and integrate crucial local representations from each modality, while the latter ensures the preservation of global information during the fusion process. We evaluate GLoMo on various datasets, encompassing tasks in multimodal sentiment analysis, multimodal humor detection, and multimodal emotion recognition. Extensive experiments demonstrate that GLoMo outperforms existing state-of-the-art models, validating the effectiveness of our proposed framework. Our code is publicly available at https://github.com/YetZzzzzz/GLoMo.

Recommendation systems, widely adopted in social networks, personalize user experiences through advanced technologies such as Reinforcement Learning (RL), known for producing high-performance, list- wise recommendations. However, RL-based recommendation methods exhibit biases, specifically: 1) Online bias, which stems from a complex real-worldonline policycomposed of various rules and models rather than a single policy; 2) Training bias, a distributional shift resulting from differences between thetarget policyand thebehavior policy. To address these issues, we introduce a novel framework named Contrastive Imitation Proximal Policy Optimization (CIPPO) for recommendation based on RL. This approach leverages extensively labeled feedback data and incorporates a Masked Imitation Network (MIN) that closely emulates the online policy, thus reducing discrepancies between online and offline environments. Additionally, the clipping function in Proximal Policy Optimization, combined with a specially designed contrastive module, effectively reduces the distributional shift between the behavior and target policies. We conduct offline and online experiments to show the improvements of CIPPO, providing details including ablation tests and parameter analysis to validate the effectiveness and robustness. CIPPO gains 12.79% on ACN and in WeChat Top Stories, a large media platform with over 50 million users.

Social media popularity prediction aims to predict future interaction or attractiveness of new posts. However, in most existing works, there is a notable deficiency in the effective treatment of numerical features. Despite their significant potential to provide ample information, these features are often inadequately processed, leading to insufficiency of information acquirement. In this paper, we introduce a method, named Double-Fine-Tuning Multi-Objective Vision-and-Language Transformer (DFT-MOVLT). To supplement the information in vision-and-language pre-training (VLP), we propose compound text, which is concatenated by numerical data and text. Furthermore, during VLP, a transformer is trained using 3 objectives to ensure thorough feature extraction. Finally, for more generalized prediction, we fine-tune 2 models using different training ways and ensemble them. To evaluate the effectiveness of each mechanism adopted in the proposed method, we conduct an array of ablation experiments. Our team achieve the 3rd place in Social Media Prediction (SMP) Challenge 2023.

Invisible watermarking is an essential measure for media publishers to declare ownership of their content, in the case of minimizing the impact on the viewing experience. In dealing with active attacks such as noise attacks, filtering attacks, geometric attacks, and lossy compression attacks, existing research still has great limitations. In this paper, through probabilistically superposition of "perturbed watermark obtain by maximum voting method" and "determined Bernoulli distribution" to restore the real embedded watermark, is used to deal with complex attack situations. Specifically, the special embedded watermark is obtained by sampling from the n-fold Bernoulli distribution with parameter p. Secondly, the HAAR wavelet transform is performed on the YUV channel of the video fixed-interval image to extract its low-pass component. Then Discrete Cosine Transform is performed to obtain its frequency domain representation. The watermark information is embedded into the frequency domain representation's singular to realize the embedded invisible watermark of video. The watermark bit of every block is determined by the maximum voting method for the disturbed watermark that performs DCT and SVD operations on the low-pass component of YUV channels. At this time, the determined Bernoulli distribution is probabilistically superimposed on the watermark information to guarantee distribution consistency. Finally, mean value operation and cluster processing are performed on the watermark information to reduce volatility. Experiments show that the method proposed in this paper has apparent advantages in solving Invisible video watermarking. With a PSNR of 41 and a corresponding BAR of 0.86, our team achieves 3rd place on the leaderboard of the Invisible Video Watermark Challenge 2023.

Power demand forecasting plays a significant role in the operation of power plants and utility companies. For data privacy, federated learning (FL) is widely adopted to aggregate local models of utility companies to a global model with very few data leaks. However, defects such as malicious updates, poisoning attacks, and low-quality data, may exist in multiple FL processes. As the general resistance to various defects is not considered by most FL approaches, a design with strong generalization is strongly needed. In this paper, we adopt DEfect-AwaRe federated soft actor-critic (DearFSAC), which dynamically assigns weights to FL's local models according to their quality. For fast and stable convergence, a deep neural network based on auto-encoder is designed for model quality evaluation and dimension reduction. Then, a deep reinforcement learning (DRL) algorithm soft actor-critic (SAC) is adopted to achieve the optimal weights assignment, considering SAC's near-optimum and sufficient exploration. We conduct simulations on power consumption data in real world. The results show that our approach performs well no matter if there exist defects or not.

Social media is an indispensable part of modern life, and social media popularity prediction (SMPP) plays a vital role in practice. In current work, the inconsistency of words in labels and titles, user feature transformation, etc have not been well noticed. In this paper, we propose a novel approach named Title-and-Tag Contrastive Vision-and-Language Transformer (TTC-VLT), combining two pre-trained vision and language transformers and other two dense feature parts for this prediction task. On one hand, in order to learn the differences between titles and tags, we design title-tag contrastive learning for title-visual and tag-visual, which separately extracts multimodal information from two types of text. On the other hand, user identification features are transformed to embedding vectors to capture user attribute details. From the extensive experiments, our approach outperforms the other methods on the social media prediction dataset. Our team achieve the 2nd place on the leader board of the Social Media Prediction Challenge 2022.

Weakly supervised anomaly detection aims at learning an anomaly detector from a limited amount of labeled data and abundant unlabeled data. Recent works build deep neural networks for anomaly detection by discriminatively mapping the normal samples and abnormal samples to different regions in the feature space or fitting different distributions. However, due to the limited number of annotated anomaly samples, directly training networks with the discriminative loss may not be sufficient. To overcome this issue, this article proposes a novel strategy to transform the input data into a more meaningful representation that could be used for anomaly detection. Specifically, we leverage an autoencoder to encode the input data and utilize three factors, hidden representation, reconstruction residual vector, and reconstruction error, as the new representation for the input data. This representation amounts to encode a test sample with its projection on the training data manifold, its direction to its projection, and its distance to its projection. In addition to this encoding, we also propose a novel network architecture to seamlessly incorporate those three factors. From our extensive experiments, the benefits of the proposed strategy are clearly demonstrated by its superior performance over the competitive methods. Code is available at: https://github.com/yj-zhou/Feature_Encoding_with_AutoEncoders_for_Weakly-supervised_Anomaly_Detection.

To support multifarious vehicular applications, content sharing among vehicles or between vehicles and infrastructures can enable efficient service provisioning. In this work, we investigate joint communication, caching, and computing (3C) resource allocation to support efficient content sharing between content providers (CPs) and content requesters (CRs) in vehicular networks. To tackle the high complexity of joint 3C resource allocation, we decouple the problem into a long-term content caching strategy to allocate caching resources plus a method for short-term CP-CR pairing and corresponding communication-computing resource allocation. Specifically, a popularity and social similarity (P-SS) based caching strategy is proposed by incorporating both physical and social information. As selfish CRs may refuse to reveal their quality of service (QoS) requirements to the CPs and lead to the information asymmetry, we adopt contract theory to allocate communication and computing resources for each potential CR-CP pair. We then propose a stable-matching based algorithm to match CPs and CRs for efficient content sharing. Simulation results verify that the proposed scheme can effectively solve the problem with low complexity.

The social media prediction task is aiming at predicting content popularity which includes social multimedia data such as photos, videos, and news. The task can not only help make better decisions for recommendation, but also reveals the public attention from evolutionary social systems. In this paper, we propose a novel approach named curriculum learning for wide multimedia-based transformer with graph target detection(CL-WMTG). The curriculum learning is designed for the transformer to improve the efficiency of model convergence. The mechanism of wide multimedia-based transformer is to make the model capable of learning cross information from text, pictures and other features(e.g. categories, location). Moreover, the graph target detection part can extract different features in the picture by pretrained model and reconstruct the features with a homogeneous graph network. We achieved third place in the SMP Challenge 2020.

Recent detection method based on machine learning demonstrates significant advantages against varieties of network attacks, and has been widely deployed in cloud applications. However, novel attacks such as Advanced Persistent Threats (APTs) could evade detection of the intrusion detection system, which may lead to serious data leakage in cloud. Existing methods studied the countermeasures to defend against evasion attacks. However, a cloud service provider (CSP) also have to balance between its expect revenue and the security risk of system with limited resources. In this paper, we present the CSP's optimal strategy for effective and safety operation, in which the CSP decides the size of users that the cloud service will provide and whether enhanced countermeasures will be conducted for discovering the possible evasion attacks. While the CSP tries to optimize its profit by carefully making a two-step decision of the defense plan and service scale, the attacker considers its expected revenue to launch evasion attacks or not. To obtain insights of such a highly coupled system, we consider a system with one CSP and one attacker with two attack choices of whether to launch an evasion attack. We propose a two-stage Stackelberg game, in which the CSP acts as the leader who decides the defense plan and service scale in Stage I, and the attacker acts as the follower that determines whether to make evasion attacks in Stage II. We derive the Nash Equilibrium by analyzing the attacker's choices under different scenarios that the CSP selects. Then, we provide the CSP's optimal strategies to maximize its revenue. The simulation results help to better understand the CSP's optimal solutions under different situations.

In the era of the Internet of Things (IoT), an immense amount of data is generated from numerous sensors and devices. Data as a service (DaaS) represents a new market whose time has come, and DaaS-based businesses are emerging quickly. Businesses across sectors begin seeing their data not only as fundamentally valuable but economically viable to distribute. Due to the exponential growth of the DaaS market, the current pricing models gradually become less suitable for the selling of data sets. A more sophisticated pricing strategy is needed to unlock the value of that data for the data vendor's (DV's) revenue growth and their customers' benefits such as online service providers (SPs). In this article, we aim to maximize the DV's profits by designing a mixed sales mechanism, which allows the DV to sell data sets separately or bundled. Particularly, we apply a multidimensional adverse selection model from contract theory to model the data set trading between DVs and SPs. The DV's surplus maximization problem is solved in the single-product case first, then extended to the multiproduct case. Furthermore, the analysis of the solution of the pricing strategy in single-product and multiproduct cases is provided. Finally, the simulation results show that the proposed pricing model can improve the DV's profits efficiently.

Video relevance computation is one of the most important tasks for the personalized online streaming service. Given the relevance of videos and viewer feedbacks, the system can provide personalized recommendations, which helps viewers discover more contents of interest in most online services. However, the computation of a video relevance table is based on viewers' implicit feedbacks such as watch and search history, which perform poorly for newly added "cold-start'' videos. Facing the cold start problem, we introduce a recommendation method with Bidirectional Encoder Representations from Transformers, which considers the continuity of ordered watching plan and trained the sequence of path from start to end named Bert4Movie. What's more, we propose a method named Movie2Vec to represent the videos in a different way. Our method has been used in our solutions of Content-based Video Relevance Prediction Challenge and got a significant improvement in the AUC.

Recently, mobile social networks (MSNs) has been playing an increasingly large proportion in people's daily life, and consequently attracted enormous amount of researches in this area, including network data collection, user behavior analysis and so on. Many of them are based on the community structure of the MSNs, the detection of which has attracted academic attention. In this paper, we propose an inter-community detection scheme under the framework of the emerging compressive sensing (CS) over graphs. Firstly, we extract the social structure among users by calculating the probability of two users encountering with each other. Then, the proposed scheme utilizes the encounter probability and the edge-clustering coefficient to define a novel additive property to make the CS algorithm applicable. Simulation results demonstrate that the proposed detection scheme can detect inter- community links more accurately than the conventional random walk based method.

As a cluster between Internet of Things (IoT) and mobile social networks (MSNs), social IoT allows close interaction between humans and things. Many applications of IoT and MSN are based on the community structure to realize efficient services, and thus the detection of the community structure has become a key problem and attracted much academic attention. In this paper, a compressive sensing (CS) over graphs-based intercommunity detection scheme is proposed for the social IoT. By exploiting the probability of two nodes encountering each other, the encounter probability and the edge-clustering coefficient are utilized to define a novel metric for each connection and construct the measurement matrix. Then a CS-based detection algorithm is proposed to detect the intercommunity links. Moreover, the edge-clustering coefficient is exploited as the prior information to further improve accuracy and reduce complexity. Simulations show that the proposed detection scheme outperforms the conventional random walk-based scheme in the social IoT.

Subway short-term ridership forecasting plays an important role in intelligent transportation systems. However, limited efforts have been made to forecast the subway short-term ridership, accounting for dynamic volatility. The traditional forecasting methods can only provide point values that are unable to offer enough information on the volatility/uncertainty of the forecasting results. To fill this gap, the aim of this paper is to incorporate the dynamic volatility into the subway short-term ridership forecasting process that not only generates the expected value of the short-term ridership but also obtains the prediction interval. Four kinds of the integrated ARIMA and GARCH models are constructed to model the mean part and volatility part of the short-term ridership. The performance of the proposed method is investigated with the real subway short-term ridership data from three stations in Beijing. The model results show that the proposed model outperforms the traditional model for all three stations. The hybrid model can significantly improving the reliability of the predicted point value by reducing the mean prediction interval length of the ridership, and improve the prediction interval coverage probability. Considering the different traffic patterns between weekday and weekend, the short-term ridership is also modeled, respectively. This paper can help management understand the dynamic volatility of the subway short-term ridership, and have the potential to disseminate more reliable subway information to travelers through the information systems.

In current wireless communication systems, there is a rapid development of location based services, which will play an essential role in the future 5G networks. One key feature in providing the service is the mobile crowdsourcing in which a central cloud node denoted as the principal collects location based data from a large group of users. In this paper, we investigate the problem of how to provide continuous incentives based on user's performances to encourage users' participation in the crowdsourcing, which can be referred to the moral hazard problem in the contract theory. We not only propose the one-dimensional performance-reward related contract, but also extend this basic model into the multi-dimensional contract. First, an incentive contract which rewards users by evaluating their performances from multiple dimensions is proposed. Then, the utility maximization problem of the principal in both one-dimension and multi-dimension are formulated. Furthermore, we detailed the analysis of the multi-dimensional contract to allocate incentives. Finally, we use the numerical results to analyze the optimal reward package, and compare the principal's utility under the different incentive mechanisms. Results demonstrate that by using the proposed incentive mechanism, the principal successfully maximizes the utilities, and the users obtain continuous incentives to participate in the crowdsourcing activity.

Cooperative content delivery using multiple air interfaces (CCDMI) is a powerful solution to mitigate congestion in cellular networks. In CCDMI, the operator distributes content to selected users that further distribute it locally among its nearby users. However, a user that is capable of contributing to CCDMI might act selfishly and refuse to participate. Although the operator can encourage user participation by offering incentives, it has incomplete information about the users' willingness to participate. In order to overcome this problem of adverse selection in CCDMI, we propose two contract-based methods under information asymmetry. In both methods, the operator designs a performance-based contract set for the users that are capable of local content distribution. Using a mathematical analysis, we show that the optimal contract under information asymmetry achieves close to optimal utility for the users and the operator, compared with the information symmetry case. Moreover, the users with high willingness to participate get positive utility and the users with low willingness get zero utility. Hence, by assigning contracts, the operator can motivate user participation, despite the information asymmetry between them. Our results verify that the proposed methods improve the system performance in terms of the utility of the operator and the users.

Wireless virtualization has emerged as a promising technology to support a diverse multitude of wireless networks due to their efficient exploitation of network resources. In a mobile virtualized network, a network operator (NO) benefits from dynamically leasing its physical resources to multiple service providers (SPs). The SPs then make profits from offering wireless services to their end users. A key challenge for the NO is to captivate the SPs into the leasing and efficiently allocate its wireless resources. To achieve the maximum profit for the NO, specialized trading mechanism needs to be designed, especially when the NO is unable to access SPs' private information. This paper proposes a framework of contract-based allocation for efficient spectrum leasing under incomplete information. First, a general system model is developed with one NO and multiple SPs engaged in an exchange for spectrum resources and money. The NO provides spectrum leases in both downlink and uplink at different multiplexing gains. Optimal trading contracts are then derived to maximize the total utility at the NO, while satisfying the SPs' requirements for the trade. Numerical results confirm our analyses on the optimal contracts and their benefits, including higher utilities and lower spectrum usages for the NO.

Nowadays, cloud computing has been identified as new opportunities for migrating to the expected agility, reuse, and adaptive capabilities that can support the ever changing IT trends, requirements and environments. However, for benefits of their own, there are various motivations for cloud service providers (CSPs) to behave unfaithfully toward cloud customers, such as invading data privacy and providing inaccurate data processing results to customers. On the other hand, constrained by computation resources and capabilities, customers mostly cannot independently process big data and perform verification on correctness and completeness. Under this information asymmetry, how to ensure the correctness and completeness of data processing, computation, and mining becomes a practically crucial issue that greatly impacts the widely deployment of cloud computing, as well as the future Internet. Though verifiable computing provides us with approaches to verify the computation results. There is hardly an approach that can ensure a 100% detection accuracy. In this paper, the concept of using penalty to decrease CSP's incentive to cheat is proposed from a contract theoretical aspect, where cloud service customers can resort to an external auditor to verify the computation results when needed. Then, an auditing contract which consists of penalty and service fee is proposed so that the CSP will be fined when being found out cheating. Subsequently, the problem is formulated to solve the efficient contract under incomplete information market. Furthermore, we give the simulation results to show the audit contract improve the efficiency of the system.

With the wide adoption of smart mobile devices, there is a rapid development of location-based services. One key feature of supporting a pleasant/excellent service is the access to adequate and comprehensive data, which can be obtained by mobile crowdsourcing. The main challenge in crowdsourcing is how the service provider (principal) incentivizes a large group of mobile users to participate. In this paper, we investigate the problem of designing a crowdsourcing tournament to maximize the principal's utility in crowdsourcing and provide continuous incentives for users by rewarding them based on the rank achieved. First, we model the user's utility of reward from achieving one of the winning ranks in the tournament. Then, the utility maximization problem of the principal is formulated, under the constraint that the user maximizes its own utility by choosing the optimal effort in the crowdsourcing tournament. Finally, we present numerical results to show the parameters' impact on the tournament design and compare the system performance under the different proposed incentive mechanisms. We show that by using the tournament, the principal successfully maximizes the utilities, and users obtain the continuous incentives to participate in the crowdsourcing activity.

In cognitive radio networks (CRNs), spectrum trading is an efficient way for secondary users (SUs) to achieve dynamic spectrum access and to bring economic benefits for the primary users (PUs). Existing methods require full payment from SU, which blocked many potential “buyers,” and thus limited the PU's expected income. To better improve PUs' revenue from spectrum trading in a CRN, we introduce a financing contract, which is similar to a sealed non-cash auction that allows SU to do financing. Unlike previous mechanism designs in CRN, the financing contract allows the SU to only pay part of the total amount when the contract is signed, known as the down payment. Then, after the spectrum is released and utilized, the SU pays the rest of payment, known as the installment payment, from the revenue generated by utilizing the spectrum. The way the financing contract carries out and the sealed non-cash auction works similarly. Thus, contract theory is employed here as the mathematical framework to solve the non-cash auction problem and form mutually beneficial relationships between PUs and SUs. As the PU may not have the full acknowledgment of the SU's transmission status, the problems of adverse selection and moral hazard arise in the two scenarios, respectively. Therefore, a joint adverse selection and moral hazard model is considered here. In particular, we present three situations when either or both adverse selection and moral hazard are present during the trading. Furthermore, both discrete and continuous models are provided in this paper. Through simulations, we show that the adverse selection and moral hazard cases serve as the upper and lower bounds of the general case where both problems are present.

Wireless network virtualization has been advocated as one of the most promising technologies to provide multifarious services and applications for the future Internet by enabling multiple isolated virtual wireless networks to coexist and share the same physical wireless resources. Based on the multiple concurrent virtual wireless networks running on the shared physical substrate, service providers can independently manage and deploy different end-users services. This paper proposes a new formulation for bandwidth allocation and routing problem for multiple virtual wireless networks that operate on top of a single substrate network to minimize the operation cost of the substrate network. We also propose a preventive traffic disruption model for virtual wireless networks to minimize the amount of traffic that service providers have to reduce when substrate links fail by incorporating $\ell _{1}$ -norm into the objective function. Due to the large number of constraints in both normal state and link failure states, the formulated problem becomes a large-scale optimization problem and is very challenging to solve using the centralized computational method. Therefore, we propose the decomposition algorithms using the alternating direction method of multipliers that can be implemented in a parallel and distributed fashion. The simulation results demonstrate the computational efficiency of our proposed algorithms as well as the advantage of the formulated model in ensuring the minimal amount of traffic disruption when substrate links fail.

Wireless network virtualization has emerged as a promising technology to provide multifarious services and applications for future wireless networks due to its efficient exploitation of network resources. In a mobile virtual network (MVN), a network operator (NO) benefits from leasing physical resources, such as subcarriers, to multiple service providers (SPs), whom then make profits from offering certain wireless services to end users. To achieve the maximum profit for the NO, or the highest efficiency of a MVN, specialized trading mechanism needs to be designed when the NO is unable to access SPs' private information about cost. In this paper, we tackle the problem of efficient trading of subcarriers under incomplete information from the SPs to the NO. First, a general system model is developed with one NO and multiple SPs engaged in a trading relation to exchange spectrum resources and money. Subsequently, an optimal trading contract is derived to maximize the total utility at the NO, while maintaining the requirements by the SPs in the trading process. Simulation results confirm our analysis on the optimal contract and its benefits.

Wireless network virtualization has emerged as a promising technology to provide a variety of services and applications for future wireless network as by enabling a more effective exploitation of network resources. In a mobile virtual network (MVN), both infrastructure provider (InP) and service provider (SP) must have a complementary relationship, as their revenues are mutually dependent. The trading of resources and services between the InP and SP is usually a long-term supply contract, and details of trades are left to be specified in the future. Thus, the returns of the InP and SP depend on their bargaining positions, ex post, and investments, ex ante. As a result, the InP and SP may hesitate to have specific investment, since it may put them at a risk of no return. In this paper, the problems of determining how the ownership of the resources affect the InP and SP's incentives to invest and how to choose the most efficient investments in an MVN are studied. First, a general system model is developed in multiple InPs and SPs engaged in a complementary relationship to exchange multiple physical and virtual resources. Subsequently, for this formulated problem, the optimal investments are derived. Furthermore, we give detailed analysis of a special case and shed light on the problem of ownership and investment efficiency by answering the question on whether the ownership of resources should be integrated or operated separately by the SP and InP. Simulation results assess the parameters that affect the efficiency of investment through simulations.

LTE, as the advanced mobile telecommunication technology, is serving heavy mobile broadband traffic nowadays. Motivated by the potential boost in performance of LTE utilizing the unlicensed spectrum, significant efforts have been devoted into the commonly referred LTE-Unlicensed technique. In this work, we investigate the carrier aggregation of licensed and unlicensed spectrum by deploying micro-cell base stations, which have access to the unlicensed spectrum, to provide cellular users a more reliable and efficient transmission. We tackle the unlicensed resource allocation problem by modeling it as a student-project allocation matching game. In addition, a postmatching procedure of resource re- allocation is introduced to guarantee unlicensed users' quality of service (QoS), as well as the system-wide stability. The simulation evaluation shows the effectiveness and efficiency of our proposed matching-based approach.

With the wide adoption of smart mobile devices, there is rapid development of location based services. One key feature of supporting a pleasant/excellent service is the access to adequate and comprehensive data, which can be obtained by mobile crowdsourcing. The main challenge in crowdsourcing is how the service provider (principal) incentivize a large group of mobile users to participate. In this paper, we investigate the problem of designing a tournament to provide continuous incentives for users by rewarding them based on the rank achieved in crowdsourcing. First, we model the user's utility of reward from achieving one of the winning ranks in the tournament. Then, the utility maximization problem of the principal is formulated, under the constraint that the user maximizes its own utility by choosing the optimal effort in the crowdsourcing tournament. Furthermore, we show that, the tournament can approximate the optimal contract under full information by step function. Finally, we present numerical results to compare the system performance under the different proposed incentive mechanisms; we show that by using the tournament, the users obtain the continuous incentives to participate in the crowdsourcing activity.

As a support to the backend storage, the content caching technique is of great importance to online social networks (e.g., Facebook), in reducing the request service latency and improving user satisfaction. However, the limited caching capacity and booming user data pose great challenges for the content-cache allocation. In this paper, we propose a three-layer content caching model, and focus on how to efficiently allocation contents to caches in order to minimize the overall service latency. We try to tackle this issue by utilizing both centralized Mix Integer Linear Programming (MILP) optimization and by modeling it as a distributed student admission (SA) stable matching problem. In the SA model, we leverage the resident-oriented Gale-Shapley (RGS) algorithm to yield a stable matching between contents and cache centers. We compare the performance between the centralized and distributed algorithms in terms of system welfare and computation analysis. Through numerical results, we prove the effectiveness of our proposed methods.

With the widely adoption of smart mobile devices, there is a rapidly development of location based services. One key feature in providing the service is the crowdsourcing in which the principal obtains essential data from a large group of users, and inversely sharing the data based service with everyone for free. In this paper, we investigate the problem of how to provide continuous incentives for users to participate in the crowdsourcing activity, which can be referred to the moral hazard problem in the contract theory. First, a performance related incentive mechanism is proposed. Then, the utility maximization problem of the principal is formulated, under the constraint that each user maximizes its own utility by choosing the optimal effort in the crowdsourcing activity. Finally, the numerical results show that by using the proposed incentive mechanism, the users obtains the continuous incentives to participate in the crowdsourcing activity, and the principal successfully maximize the utilities.

In device-to-device (D2D) communication, mobile users communicate directly without going through the base station. D2D commutation has the advantage of improving spectrum efficiency. But the interference introduced by resource sharing of D2D has become a significant challenge. In this paper, we try to optimize the system throughput while simultaneously meeting the quality of service (QoS) requirements for both D2D users and cellular users (CUs). We implement matching theory to solve the resource allocation problem. We utilize two efficient stable matching algorithms to optimize the social welfare while ensuring the network stability. More importantly, we introduce the idea of cheating in matching to further improve D2D users' throughput. It is proven that the cheating mechanism benefits a subset of D2D users without hurting the performance of the rest. Through the simulation results, we demonstrate the effectiveness of both the stable matching and cheating algorithms in terms of improving both D2D users and the overall throughput in D2D communications.

Device-to-device (D2D) communication is viewed as one promising technology for boosting the capacity of wireless networks and the efficiency of resource management. D2D communication heavily depends on the participation of users in sharing contents. Thus, it is imperative to introduce new incentive mechanisms to motivate such user involvement. In this paper, a contract-theoretic approach is proposed to solve the problem of providing incentives for D2D communication in cellular networks. First, using the framework of contract theory, the users' preferences toward D2D communication are classified into a finite number of types, and the service trading between the base station and users is properly modeled. Next, necessary and sufficient conditions are derived to provide incentives for users' engagement in D2D communication. Finally, our analysis is extended to the case in which there is a continuum of users. Simulation results show that the contract can effectively incentivize users' participation, and increase capacity of the cellular network than the other mechanisms.

Uncertainty is often associated with travel-time prediction. Traditional point prediction methods only provide point values that are unable to offer enough information on the reliability of prediction results. The recent development of statistical volatility models has given us an effective way to capture uncertainties in data. Generalized autoregressive conditional heteroskedasticity (GARCH) models have been widely used in transportation systems as a way to account for this uncertainty by providing more accurate prediction intervals. However, a GARCH model arguably does not consider the trend and seasonality in data. If there is a trend or seasonality, the performance of the GARCH model may be affected. In the context of travel-time prediction, this paper proposes two component GARCH models that are able to model trend and seasonal components through decomposition. The travel-time data obtained along a freeway corridor in Houston, TX, USA, were used to empirically test the performance of the proposed models. The study results indicate that the proposed models perform well when capturing uncertainties associated with travel-time prediction.

Device-to-device (D2D) communication is seen as a major technology to overcome the imminent wireless capacity crunch and to enable new application services. In this paper, a novel social-aware approach for optimizing D2D communication by exploiting two layers, namely the social network layer and the physical wireless network layer, is proposed. In particular, the physical layer D2D network is captured via the users' encounter histories. Subsequently, an approach, based on the so-called Indian Buffet Process, is proposed to model the distribution of contents in the users' online social networks. Given the social relations collected by the base station, a new algorithm for optimizing the traffic offloading process in D2D communications is developed. In addition, the Chernoff bound and approximated cumulative distribution function (cdf) of the offloaded traffic are derived and the validity of the bound and cdf is proven. Simulation results based on real traces demonstrate the effectiveness of our model and show that the proposed approach can offload the network's traffic successfully.

In device-to-device (D2D) communication, mobile users communicate directly without going through the base station. D2D commutation has the advantage of improving spectrum efficiency. But the interference introduced by resource sharing of D2D has become a significant challenge. In this paper, we try to maximize the system throughput and simultaneously meet the Quality of Service (QoS) requirement for both D2D users and cellular users (CUs). We solve the optimization problem by solving the bipartite two-sided matching problem under preferences between the admitted sets of D2D users and CUs. Second and more importantly, we introduce the idea of cheating in matching to further improve some D2D users' system throughputs. We construct a coalition strategy to implement the cheating idea. Finally, from the simulation results, we demonstrate the effectiveness of cheating in matching for D2D pairs.

Numerous applications of wireless sensor networks (WSNs) in harsh terrains are constrained by the sensors' battery-power and face the difficulties of data collection. In this paper, we propose to exploit wireless power transfer technology to replenish the energy of sensor clusters and develop an efficient data collection scheme for those wireless rechargeable senor clusters deployed in harsh terrains. In view of the harsh terrains, we employ unmanned aerial vehicles (UAVs) to travel to the sites of sensor clusters, collect data, and recharge the sensors in corresponding clusters. With joint consideration of data collection characteristics, wireless power transfer features and travel time, we mathematically formulate the data collection in rechargeable WSNs into an optimization problem with the objective of maximizing data collection utility. Based on the matching theory, we also develop a one side matching algorithm and a greedy algorithm to solve the problem in distributed manner. Through simulations, we show that UAVs are not always matched with nearest sensor clusters, the solution of the proposed greedy algorithm is optimal, and the sensed data can be efficiently collected.

In cognitive radio (CR) networks, due to the uncertainty of primary users' (PUs) traffic, there are vacant bandwidths that can be accessed by secondary users (SUs). Thus, this provides mutual benefit opportunities for both PUs and SUs. In this paper, the objective is to select the optimal channel utilization for maximizing the system utility. To solve the problem, we employ the matching theory as the mathematical framework to form mutually beneficial relationships between PUs and SUs. In particular, we propose a distributed matching algorithm for many-to-one matching considering both SUs' and PUs' characteristics. Simulation results show that the proposed distributed algorithm achieves a performance close to the centralized method, and outperforms the greedy allocation in different scenarios.