Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Randall L. BarbourSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works572 from public data
- Structural Deep Network Embedding2,836
This paper proposes a Structural Deep Network Embedding method, namely SDNE, which first proposes a semi-supervised deep model, which has multiple layers of non-linear functions, thereby being able to capture the highly non- linear network structure and exploits the first-order and second-order proximity jointly to preserve the network structure.
- Deep Learning on Graphs: A Survey1,644
This survey comprehensively review the different types of deep learning methods on graphs by dividing the existing methods into five categories based on their model architectures and training strategies: graph recurrent neural networks, graph convolutional networks,graph autoencoders, graph reinforcement learning, and graph adversarial methods.
- Asymmetric Transitivity Preserving Graph Embedding1,333
A novel graph embedding algorithm, High-Order Proximity preserved Embedding (HOPE for short), is developed, which is scalable to preserve high-order proximities of large scale graphs and capable of capturing the asymmetric transitivity.
- A Survey on Network Embedding1,269
This survey focuses on categorizing and then reviewing the current development on network embedding methods, and point out its future research directions, covering the structure- and property-preserving network embeding methods, the network embedded methods with side information, and the advanced information preserving network embedting methods.
- 1,073
- Community Preserving Network Embedding935
A novel Modularized Nonnegative Matrix Factorization (M-NMF) model is proposed to incorporate the community structure into network embedding and jointly optimize NMF based representation learning model and modularity based community detection model in a unified framework, which enables the learned representations of nodes to preserve both of the microscopic and community structures.
- Robust Graph Convolutional Networks Against Adversarial Attacks580
Robust GCN (RGCN), a novel model that "fortifies'' GCNs against adversarial attacks by adopting Gaussian distributions as the hidden representations of nodes in each convolutional layer, which can automatically absorb the effects of adversarial changes in the variances of the Gaussian distribution.
- 473
- Depression Detection via Harvesting Social Media: A Multimodal Dictionary Learning Solution404
A multimodal depressive dictionary learning model is proposed to detect the depressed users on Twitter and a series of experiments are conducted to validate this model, which outperforms (+3% to +10%) several baselines.
- To Find Where You Talk: Temporal Sentence Localization in Video with Attention Based Location Regression388
A novel attention based location regression network is designed to predict the temporal coordinates of sentence query from the previous attention and a multi-modal co-attention mechanism is introduced to generate not only video attention which reflects the global video structure, but also sentence attention which highlights the crucial details for temporal localization.
- Disentangled Graph Convolutional Networks387
This paper introduces the disentangled graph convolutional network (DisenGCN) to learn disentangled node representations and proposes a novel neighborhood routing mechanism, which is capable of dynamically identifying the latent factor that may have caused the edge between a node and one of its neighbors and accordingly assigning the neighbor to a channel that extracts and convolutes features to that factor.
- VTimeLLM: Empower LLM to Grasp Video Moments371
Benefits from the fine-grained temporal understanding of the videos further enable VTimeLLM to beat existing Video LLMs in video di-alogue benchmark, showing its superior cross-modal understanding and reasoning abilities.
- Social contextual recommendation317
This paper investigates social recommendation on the basis of psychology and sociology studies, which exhibit two important factors: individual preference and interpersonal influence, and proposes a novel probabilistic matrix factorization method to fuse them in latent spaces.
- Semantic Conditioned Dynamic Modulation for Temporal Sentence Grounding in Videos303
A novel semantic conditioned dynamic modulation mechanism, which leverages the sentence semantics to modulate the temporal convolution operations for better correlating and composing the sentence-relevant video contents over time.
- BeautyGAN283
A dual input/output Generative Adversarial Network that enables the network to learn translation on instance-level through unsupervised adversarial learning, and could generate visually pleasant makeup faces and accurate transferring results.
- Disentangled Self-Supervision in Sequential Recommenders271
This paper proposes a sequence-to-sequence (seq2seq) training strategy based on latent self-supervision and disentanglement, which performs self- supervision in the latent space, instead of reconstructing the items in the future sequence individually.
- 268
- Disentangled Representation Learning260
This article comprehensively investigates DRL from various aspects including motivations, definitions, methodologies, evaluations, applications, and model designs, and points out challenges in DRL as well as potential research directions deserving future investigations.
- 249
- 248
- A Comprehensive Survey on Deep Clustering: Taxonomy, Challenges, and Future Directions247
A novel taxonomy of deep clustering is introduced based on how two fundamental components: the representation learning module and the clustering module interact, specifically through multistage, generative, iterative, and simultaneous approaches.
- Structural Deep Embedding for Hyper-Networks216
It is theoretically prove that any linear similarity metric in embedding space commonly used in existing methods cannot maintain the indecomposibility property in hyper-networks, and thus a new deep model is proposed to realize a non-linear tuplewise similarity function while preserving both local and global proximities in the formedembedding space.
- Arbitrary-Order Proximity Preserved Network Embedding210
The eigen-decomposition reweighting theorem is theoretically proved, revealing the intrinsic relationship between proximities of different orders and proposed AROPE (arbitrary-order proximity preserved embedding), a novel network embedding method based on SVD framework.
- AVQA: A Dataset for Audio-Visual Question Answering on Videos199
This paper introduces AVQA, a new audio-visual question answering dataset on videos in real-life scenarios and proposes a Hierarchical Audio-Visual Fusing module to model multiple semantic correlations among audio, visual, and text modalities and conduct ablation studies to analyze the role of different modalities on the authors' datasets.
- Modelling and performance analysis of the distributed scheduler in IEEE 802.16 mesh mode188
This work develops a stochastic model for the distributed scheduler of the mesh mode of the IEEE 802.16 protocol and analyzes the scheduler performance under various conditions, and the analytical results match very well with the ns-2 simulation results.
- 185
- A Restricted Black-Box Adversarial Framework Towards Attacking Graph Embedding Models180
This paper investigates the theoretical connections between graph signal processing and graph embedding models in a principled way and forms a generalized adversarial attacker, GF-Attack, which is constructed by the graph filter and feature matrix and validated on several benchmark datasets.
- Deep Recursive Network Embedding with Regular Equivalence180
This work proposes a new approach named Deep Recursive Network Embedding (DRNE) to learn network embeddings with regular equivalence, and proposes a layer normalized LSTM to represent each node by aggregating the representations of their neighborhoods in a recursive way.
- Out-of-Distribution Generalization on Graphs: A Survey173
This paper comprehensively survey OOD generalization on graphs and presents a detailed review of recent advances in this area, including a formal problem definition of OOD generalization on graphs.
- Understanding Performance of Edge Content Caching for Mobile Video Streaming172
This paper uses real-world data sets containing 50 million trace items of nearly 2 million users viewing more than 0.3 million unique videos using mobile devices in a metropolis in China over two weeks to understand the request patterns and user behaviors in mobile video streaming, and designs an efficient caching strategy based on the measurement insights and experimentally evaluate its performance.
- JALAD: Joint Accuracy-And Latency-Aware Deep Structure Decoupling for Edge-Cloud Execution170
JALAD is proposed, a joint accuracy- and latency-aware execution framework, which decouples a deep neural network so that a part of it will run at edge devices and the other part inside the conventional cloud, while only a minimum amount of data has to be transferred between them.
- 165
- OOD-GNN: Out-of-Distribution Generalized Graph Neural Network162
This work proposes an out-of-distribution generalized graph neural network (OOD-GNN) for achieving satisfactory performance on unseen testing graphs that have different distributions with training graphs and presents a global weight estimator to learn weights for training graphs such that variables in graph representations are forced to be independent.
- Multiresolution watermarking for images and video162
It is shown that when subjected to distortion from compression or image halftoning, the corresponding watermark can still be correctly identified at each resolution (excluding the lowest one) in the wavelet domain.
- Resource allocation for multimedia streaming over the Internet155
An end-to-end transport architecture for multimedia streaming over the Internet is presented and a new multimedia streaming TCP-friendly protocol is proposed, which combines forward estimation of network conditions with information feedback control to optimally track the network conditions.
- Cascading outbreak prediction in networks140
This paper proposes Orthogonal Sparse LOgistic Regression (OSLOR) method to jointly optimize node selection and outbreak prediction, where the prediction loss are combined with an orthogonal regularizer and L1 regularizer to guarantee good prediction accuracy, as well as the sparsity and low-redundancy of selected sensors.
- 139
- Deep multimodal hashing with orthogonal regularization133
This paper proposes a novel deep multimodal hashing method, namely Deep Multimodal Hashing with Orthogonal Regularization (DMHOR), which fully exploits intra- modality and inter-modality correlations and finds that a better representation can be attained with different numbers of layers for different modalities, due to their different complexities.
- Deep Variational Network Embedding in Wasserstein Space129
The experimental results demonstrate that the proposed Deep Variational Network Embedding in Wasserstein Space (DVNE) can effectively model the uncertainty of nodes in networks, and show a substantial gain on real-world applications such as link prediction and multi-label classification compared with the state-of-the-art methods.
- 129
- Two decades of internet video streaming129
A retrospective view of the research results over the past two decades, with a focus on peer-to-peer streaming protocols and the effects of cloud computing and social media is provided.
- Power-minimized bit allocation for video communication over wireless channels127
This work investigates the relations of rate, distortion and power consumption in video coding and proposes a power-minimized bit-allocation scheme considering the processing power, for source coding and channel coding, jointly with the transmission power.
- Joint Content Replication and Request Routing for Social Video Distribution Over Cloud CDN: A Community Clustering Method124
This paper presents a social video replication and user request dispatching mechanism in the cloud content delivery network architecture to reduce the system operational cost, while guaranteeing the averaged service latency.
- A cross-layer quality-of-service mapping architecture for video delivery in wireless networks117
This paper presents a cross-layer mapping architecture for video transmission in wireless networks and describes the design and algorithms for each building block, which either builds upon or extend the state-of-the-art algorithms that were developed without much considerations of other layers.
- 115
- 114
- TIMERS: Error-Bounded SVD Restart on Dynamic Networks112
This paper proposes TIMERS, Theoretically Instructed Maximum-Error-bounded Restart of SVD, a novel approach which optimally sets the restart time in order to reduce error accumulation in time, and monitors the margin between reconstruction loss of incremental updates and the minimum loss in SVD model.
- 107
- Boosting Visual Question Answering with Context-aware Knowledge Aggregation106
The proposed KG-Aug model is capable of retrieving context-aware knowledge subgraphs given visual images and textual questions, and learning to aggregate the useful image- and question-dependent knowledge which is then utilized to boost the accuracy in answering visual questions.
- Cross-media analysis and reasoning: advances and directions104
By presenting approaches, advances, and future directions in cross-media analysis and reasoning, the goal is to draw more attention to the state-of-the-art advances in the field, and to provide technical insights by discussing the challenges and research directions in these areas.
- TIVA-KG: A Multimodal Knowledge Graph with Text, Image, Video and Audio103
The proposed TIVA-KG has coverage of up to four modalities including text, image, video, audio, and capability of triplet grounding which grounds multimodal relations to triples instead of entities, and a Quadruple Embedding Baseline model is designed to validate the necessity and efficacy of considering fourmodalities in KG.
- 102
- Automated Machine Learning on Graphs: A Survey98
This paper comprehensively survey AutoML on graphs, primarily focusing on hyper-parameter optimization (HPO) and neural architecture search (NAS) for graph machine learning.
- Dependence Guided Unsupervised Feature Selection98
This work proposes a Dependence Guided Unsupervised Feature Selection (DGUFS) method to select features and partition data in a joint manner and utilizes the learned cluster labels to fill in the information gap between original data and selected features.
- 98
- DisenBooth: Identity-Preserving Disentangled Tuning for Subject-Driven Text-to-Image Generation97
This work proposes DisenBooth, an identity-preserving disentangled tuning framework for subject-driven text-to-image generation that finetunes the pretrained diffusion model in the denoising process to preserve the subject identity and capture the identity-irrelevant information.
- Curriculum Meta-Learning for Next POI Recommendation96
A novel Curriculum Hardness Aware Meta-Learning (CHAML) framework is proposed, which incorporates hard sample mining and curriculum learning into a meta-learning paradigm and uses an easy-to-hard curriculum for the city-sampling pool to help the meta-learner converge to a better state.
- Projection-free Distributed Online Learning in Networks96
The distributed online conditional gradient algorithm is proposed, which eschews the expensive projection operation needed in its counterpart algorithms by exploiting much simpler linear optimization steps and gives a regret bound for the proposed algorithm as a function of the network size and topology.
- Smart-Phone Attacks and Defenses96
Smart-phones, interoperable between the telecom networks and the Internet, are dangerous conduits for Internet security threats to reach the telecom infrastructure and the damage caused by subverted smart-phones could range from privacy violation and identity theft to emergency call center DDoS attacks and national crises.
- Billion-Scale Network Embedding with Iterative Random Projection95
RandNE (Iterative Random Projection Network Embedding), a novel and simple billion-scale network embedding method, which proposes a Gaussian random projection approach to map the network into a low-dimensional embedding space while preserving the high-order proximities between nodes.
- Propagation-based social-aware replication for social video contents94
A propagation-based social-aware replication framework using a hybrid edge-cloud and peer-assisted architecture, namely PSAR, to serve the social video contents is proposed, which improves the replication performance and the video service quality.
- 92
- Towards Cost-Efficient Video Transcoding in Media Cloud: Insights Learned From User Viewing Patterns89
- 87
- 86
- Structurally Incoherent Low-Rank Nonnegative Matrix Factorization for Image Classification85
This paper proposes a novel nonnegative factorization method, called structurally incoherent low-rank NMF (SILR-NMF), in which they jointly consider structural incoherence and low- rank properties of data for image classification.
- A Closer Look at Temporal Sentence Grounding in Videos: Dataset and Metric83
This paper proposes to re-organize two widely-used TSGV benchmarks (ActivityNet Captions and Charades-STA) and introduces a new evaluation metric "dR@n,IoU@m'' to calibrate the basic IoU scores by penalizing on the bias-influenced moment predictions and alleviate the inflating evaluations caused by the dataset annotation biases such as overlong ground-truth moments.
- 83
- 82
- Social recommendation across multiple relational domains81
A novel Hybrid Random Walk (HRW) is proposed, which can integrate multiple heterogeneous domains including directed/undirected links, signed/unsigned links and within-domain/cross-domain links into a star-structured hybrid graph with user graph at the center and use the steady state distribution for recommendation.
- 80
- Second-order derivative-based smoothness measure for error concealment in DCT-based codecs79
Simulation results show that a weighted combination of the quadratic variation and the Laplacian operator can significantly reduce the blurring across the edges while enforcing smoothness along the edges.
- LLM4DyG: Can Large Language Models Solve Spatial-Temporal Problems on Dynamic Graphs?78
This paper proposes the LLM4DyG benchmark, which includes nine specially designed tasks considering the capability evaluation of LLMs from both temporal and spatial dimensions, and proposes Disentangled Spatial-Temporal Thoughts (DST2) for LLMs on dynamic graphs to enhance LLMs' spatial-temporal understanding abilities.
- Multimedia Big Data Computing78
The authors introduce the principal concepts for multimedia big data computing, discuss what the scientific problems and fundamental challenges are, and present methodologies and approaches from the perspectives of the multimedia life cycle and multimediabig data computing life cycle.
- 78
- 76
- Diff4Rec: Sequential Recommendation with Curriculum-scheduled Diffusion Augmentation75
The proposed Diff4Rec framework is able to effectively achieve superior performance over several strong baselines, capable of making high-quality and robust sequential recommendations, and has the promising potential to bring paradigm shift in multimedia recommendation.
- 73
- Fast Hardware-Aware Neural Architecture Search73
This paper proposes HURRICANE, an automatic hardware-aware search over a much larger search space and a two-stage search algorithm, to efficiently generate tailored models for different types of hardware, which achieve much lower latency and higher accuracy than current state-of-the-art efficient models.
- 73
- Iterative total least-squares image reconstruction algorithm for optical tomography by the conjugate gradient method71
An iterative total least-squares algorithm for computing images of the interior structure of highly scattering media by using the conjugate gradient method that is particularly suitable for very large matrices.
- A Survey on Temporal Sentence Grounding in Videos70
This survey gives a comprehensive overview for TSGV, summarizes the taxonomy of existing methods, provides a detailed description of the evaluation protocols, and in-depth discusses potential problems of current benchmarking designs and research directions for further investigations.
- DepthLGP: Learning Embeddings of Out-of-Sample Nodes in Dynamic Networks70
This work designs a high-order Laplacian Gaussian process (hLGP) to encode network properties, which permits fast and scalable inference, and designs a deep neural network to learn a nonlinear transformation from latent states of the hLGP to node embeddings.
- QoS-adpative proxy caching for multimedia streaming over the Internet70
A media-characteristic-weighted replacement policy is proposed to improve the cache hit ratio of mixed media including continuous and noncontinuous media and a network-condition- and media-quality-adaptive resource-management mechanism is introduced to dynamically re-allocate cache resource for different types of media according to their request patterns.
- Comparing apples to oranges69
This paper proposes a novel Relation-aware Heterogeneous Hashing (RaHH), which provides a general framework for generating hash codes of data entities sitting in multiple heterogeneous domains and encodes both homogeneous and heterogeneous relationships between the data entities to design hash functions with improved accuracy.
- Signed Graph Neural Network with Latent Groups61
The proposed Group Signed Graph Neural Network model for signed graph representation learning is an effective approach to analyze the complex patterns in real-world signed graphs with the co-existence of positive and negative links and adopts a relational GNN to learn node representations.
- 60
- Conditional Graph Attention Networks for Distilling and Refining Knowledge Graphs in Recommendation59
This work proposes Knowledge-aware Conditional Attention Networks (KCAN), which is an end-to-end model to incorporate knowledge graph into a recommender system and can gain both representability and personalization to achieve overall performance.
- 59
- Multimodal Continual Graph Learning with Neural Architecture Search57
This paper proposes a novel Multimodal Structure-evolving Continual Graph Learning (MSCGL) model, which continually learns both the model architecture and the corresponding parameters for Adaptive Multinodal Graph Neural Network (AdaMGNN).
- Multi-Modal Deep Analysis for Multimedia57
A deep and comprehensive overview for multi-modal analysis in multimedia, introducing two scientific research problems, data-driven correlational representation and knowledge-guided fusion for multimedia analysis, and discusses the approaches for fusing knowledge with data.
- Dynamic Heterogeneous Graph Attention Neural Architecture Search54
This paper proposes a novel Dynamic Heterogeneous Graph Attention Search (DHGAS) method, which can automatically discover the optimal DHGNN architecture and adapt to various dynamic heterogeneous graph scenarios without human guidance and significantly outperforms state-of-the-art baselines for tasks including link prediction, node classification and node regression.
- 54
- MEmoR: A Dataset for Multimodal Emotion Reasoning in Videos53
This work introduces a new dataset, MEmoR, that provides fine-grained emotion annotations for both speakers and non-speakers, and proposes an attention-based reasoning approach to model the intra-personal emotion contexts, inter- personal emotion propagation, and the personalities of different individuals.
- Adaptive Disentangled Transformer for Sequential Recommendation52
The Adaptive Disentangled Transformer (ADT) framework is proposed, which is able to adaptively determine the optimal degree of disentanglement of attention heads within different layers, and requires the independence constraint via mutual information estimation over attention heads and employing auxiliary objectives to prevent the information from collapsing into useless noise.
- Disentangled Representation Learning with Large Language Models for Text-Attributed Graphs52
The Disentangled Graph-Text Learner (DGTL) model is presented, which is able to enhance the reasoning and predicting capabilities of LLMs for TAGs and can offer natural language explanations for predictions, thereby significantly enhancing model interpretability.
- Eigen-GNN: a Graph Structure Preserving Plug-in for GNNs52
This work proposes Eigen-GNN, a simple yet effective and general plug-in module to boost GNNs ability in preserving graph structures by treating GNN’s as a type of dimensionality reduction and expanding the initialdimensionality reduction bases.
- 52
- Deep Asymmetric Transfer Network for Unbalanced Domain Adaptation52
This paper proposes a novel Deep Asymmetric Transfer Network (DATN) to address the problem of unbalanced domain adaptation and proposes an unsupervised transfer method to propagate the knowledge from a lot of unsuper supervised data by minimizing the distribution discrepancy over the unlabeled data of two domains.
- FEMA51
A Flexible Evolutionary Multi-faceted Analysis framework for both behavior prediction and pattern mining that utilizes a flexible and dynamic factorization scheme for analyzing human behavioral data sequences, which can incorporate various knowledge embedded in different object domains to alleviate the sparsity problem.
- 50
- 49
- 48
- 48
- Cross-layer QoS Support for Multimedia Delivery over Wireless Internet48
This paper presents a framework, which provides QoS support, for multimedia delivery over wireless Internet, across different layers, and proposes a cross-layer architecture, which combines the application-level, transport-layer, as well as link-layer controls, and then review recent advances in each individual component.
- Spectral Graph Attention Network with Fast Eigen-approximation46
Spectral Graph Attention Network (SpGAT) that learns representations for different frequency components regarding weighted filters and graph wavelets bases is presented that can better capture global patterns of graphs in an efficient manner with much fewer learned parameters than that of GAT.
- AutoNE46
This paper proposes a novel framework, named AutoNE, to automatically optimize the hyperparameters of a NE algorithm on massive networks, and employs a multi-start random walk strategy to sample several small sub-networks, perform each trial of configuration selection on the sampled sub-network, and design a meta-leaner to transfer the knowledge about optimal hyperparameter from the sub-nets to the original massive network.
- Embodied-R: Collaborative Framework for Activating Embodied Spatial Reasoning in Foundation Models via Reinforcement Learning45
Embodied-R, a collaborative framework combining large-scale Vision-Language Models (VLMs) for perception and small-scale Language Models (LMs) for reasoning, achieves slow-thinking capabilities with limited computational resources using Reinforcement Learning (RL).
- 45
- 44
- 44
- 44
- Advances in neural architecture search43
This paper delves into the multifaceted aspects of NAS, elaborating on its recent advances, applications, tools, benchmarks and prospective research directions.
- Hierarchical Taxonomy Aware Network Embedding43
NetHiex is proposed, a NETwork embedding model that captures the latent HIErarchical taXonomy and employs the nested Chinese restaurant process to guide the search of the most plausible hierarchical taxonomy.
- 43
- Towards Better Alignment: Training Diffusion Models with Reinforcement Learning Against Sparse Rewards42
A novel RL-based framework that addresses the sparse reward problem when training diffusion models, named B2-DiffuRL, which employs two strategies: Backward progressive training and Branch-based sampling.
- 42
- 42
- Resource allocation for cloud-based free viewpoint video rendering for mobile phones42
Experimental results demonstrate that the proposed cloud-based FVV rendering solution can substantially improve video quality on mobile devices comparing with traditional approaches.
- 42
- 41
- User interest and social influence based emotion prediction for individuals41
A novel emotion prediction method for individuals based on user interest and social influence is proposed and a simple yet efficient weight learning method in which the weights are obtained from users' behaviors is proposed.
- Intra- and Inter-Modal Curriculum for Multimodal Learning40
The proposed I2MCL, a generic Intra- and Inter-Modal Curriculum Learning framework which simultaneously considers both data difficulty and modality balance for multimodal learning, is proposed.
- 40
- Cross-Modal Dual Learning for Sentence-to-Video Generation40
The proposed CMDL model adopts a dual learning mechanism to simultaneously learn the bidirectional mappings between sentences and videos such that it is able to generate realistic videos which maintain semantic consistencies with their corresponding textual descriptions.
- Social-Sensed Image Search40
An example-based user interest representation and modeling method to integrate the user-image interest graph from social media and search results from general image search engines to rerank the images by fusing their social relevance and visual relevance.
- Allocation of layer bandwidths and FECs for video multicast over wired and wireless networks40
This work considers a multicast system in which the server adapts the bandwidth and forward-error correction code (FEC) of each layer so as to maximize the overall video quality, and proposes an efficient algorithm on FEC allocation for the base layer.
- 39
- 39
- 38
- Social Recommendation with Optimal Limited Attention38
This work proposes a novel algorithm through employing an EM-style strategy to jointly optimize users' latent preferences, optimal number of their best influential friends and the corresponding attentions, and develops a rigorous proof to guarantee the algorithm's optimality.
- 38
- 38
- 37
- Precoded OFDM with adaptive vector channel allocation for scalable video transmission over frequency-selective fading channels37
This paper proposes to enhance the precoded OFDM system with adaptive vector channel allocation to provide stronger protection against errors to more important layers in the layered bit stream structure of scalable video.
- Precoded OFDM with Adaptive Vector Channel Allocation for Scalable Video Transmission37
This paper proposes to enhance the precoded OFDM system with adaptive vector channel allocation to provide stronger protection against errors to more important layers in the layered bit stream structure of scalable video.
- 36
- 36
- 36
- Sentence Specified Dynamic Video Thumbnail Generation35
A novel graph convolved video thumbnail pointer (GTP) that leverages a sentence specified video graph convolutional network to model both the sentence-video semantic interaction and the internal video relationships incorporated with the sentence information, based on which a temporal conditioned pointer network is introduced to sequentially generate the sentence specifying video thumbnails.
- A unified framework for resource discovery and QoS-aware provider selection in ad hoc networks35
This paper identifies the challenges when basic resource discovery techniques for the Internet are used in mobile ad hoc networks and proposes a framework that provides a unified solution to the discovery of resources and QoS-aware selection of resource providers.
- 34
- Incorporating External Knowledge to Answer Open-Domain Visual Questions with Dynamic Memory Networks34
- Dynamic Mixture of Curriculum LoRA Experts for Continual Multimodal Instruction Tuning33
A novel Dynamic Mixture of Curriculum LoRA Experts (D-MoLE) method, which automatically evolves MLLM's architecture with controlled parameter budgets to continually adapt to new tasks while retaining previously learned knowledge, is proposed, the first study of continual learning for MLLMs from an architectural perspective.
- 33
- Curriculum-Listener: Consistency- and Complementarity-Aware Audio-Enhanced Temporal Sentence Grounding33
Adaptive Dual-branch Promoted Network (ADPN) is proposed, which introduces a dual-br branch pipeline capable of jointly training visual-only and audio-visual branches to simultaneously eliminate inter-modal interference and designs Text-Guided Clues Miner (TGCM) to discover crucial locating clues via considering both consistency and complementarity during audio- visual interaction guided by text semantics.
- 33
- Power up! Robust Graph Convolutional Network via Graph Powering33
This work proposes a new convolution operator that is provably robust in the spectral domain and is incorporated in the GCN architecture to improve expressivity and interpretability and proposes a robust training paradigm that encourages transferability across graphs that span a range of spatial and spectral characteristics.
- Dest-ResNet33
Dest-ResNet (Deep spatiotemporal Residual Network) is proposed for hotspot traffic speed prediction and shows a 30% relative boost over the state-of-the-art methods on real-world datasets from Baidu Map.
- 33
- Power up! Robust Graph Convolutional Network against Evasion Attacks based on Graph Powering.32
A robust learning paradigm is proposed, where the network is trained on a family of "'smoothed" graphs that span a spatial and spectral range for generalizability and the new operator is used in replacement of the classical Laplacian to construct an architecture with improved spectral robustness, expressivity and interpretability.
- 32
- Social Media Recommendation32
This chapter presents the framework of social media recommendation, with a focus on two important types of recommendations: interest-orientedsocial media recommendation and influence-oriented social mediaRecommendation, which takes both social property and content property into account.
- 32
- DisenStudio: Customized Multi-Subject Text-to-Video Generation with Disentangled Spatial Control31
The proposed DisenStudio is a novel framework that can generate text-guided videos for customized multiple subjects, given few images for each subject, and significantly outperforms existing methods in various metrics.
- 31
- Discrete Social Recommendation31
This work proposes a novel discrete social recommendation (DSR) method which learns binary codes in a unified framework for users and items, considering social information, and puts the balanced and uncorrelated constraints on the objective to ensure the learned binary codes can be informative yet compact.
- Prefetching strategy in peer-assisted social video streaming31
This work designs a user preference guided prefetching strategy to reduce video startup delays, enabling smooth playback and achieves high prefetch accuracy and short startup delay with conservative storage and bandwidth capacities at peers.
- Cost-Based Cache Replacement and Server Selection for Multimedia Proxy Across Wireless Internet31
A novel replacement algorithm for single-server and a new server-selection policy for multiple servers are proposed to improve the end-to-end performance such as throughput, media quality, and start-up latency of multimedia proxy in wireless Internet.
- Invariant Node Representation Learning under Distribution Shifts with Multiple Latent Environments30
This article proposes a novel Invariant Node representation Learning approach capable of generating invariant node representations based on the invariant patterns under distribution shifts with multiple latent environments by leveraging the invariance principle.
- 30
- Exploring the Potential of Large Language Models in Graph Generation29
This paper proposes LLM4GraphGen to explore the ability of LLMs for graph generation with systematical task designs and extensive experiments, and demonstrates that LLMs, particularly GPT-4, exhibit preliminary abilities in graph generation tasks, including rule-based and distribution-based generation.
- Disentangled Dynamic Graph Attention Network for Out-of-Distribution Sequential Recommendation29
This article proposes Disentangled Intervention-based Dynamic graph Attention networks with Invariance Promotion (I-DIDA) to handle spatio-temporal distribution shifts in sequential recommendation by discovering and utilizing invariant patterns, i.e., structures and features whose predictive abilities are stable across distribution shifts.
- Curriculum Multi-Negative Augmentation for Debiased Video Grounding29
This work proposes multiple negative augmentations in a hierarchical way, including cross-video augmentations from clip-/video-level, and self-shuffled augmentations with masks, to discourage the standard VG model's behavior of exploiting such temporal annotation biases and improve the model generalization ability.
- Partial Multi-View Outlier Detection Based on Collective Learning29
A novel Collective Learning (CL) based framework to detect outliers from partial multi-view data in a self-guided way and proposes similarity-based outlier detection to break through the dilemma that the number of clusters is unknown priori.
- 29
- 29
- Robust Scalable Video Streaming over Internet with Network-Adaptive Congestion Control and Unequal Loss Protection29
A new network-adaptive congestion control scheme for scalable video streaming in conjunction with Unequal Loss Protection (ULP) that not only dynamically adapts the bit rate to the varying network condition, but also has the error-concealment capability.
- Multi-weather Cross-view Geo-localization Using Denoising Diffusion Models28
MCGF establishes a joint optimization between image restoration and geo-localization using denoising diffusion models, and achieves competitive results for geo-localization in varying weather conditions.
- U2UData: A Large-scale Cooperative Perception Dataset for Swarm UAVs Autonomous Flight28
U2UData is presented, the first large-scale cooperative perception dataset for swarm UAVs autonomous flight and introduces two perception tasks: cooperative 3D object detection and cooperative 3D object tracking.
- RealTCD: Temporal Causal Discovery from Interventional Data with Large Language Model28
This paper develops a score-based temporal causal discovery method capable of discovering causal relations without relying on interventional targets through strategic masking and regularization and introduces LLM-guided meta-initialization to extract the meta-knowledge from textual information hidden in systems to boost the quality of discovery.
- Bilateral Correspondence Model for Words-and-Pictures Association in Multimedia-Rich Microblogs28
This paper uses a data-driven approach to analyze and modeling the associations between multimodal contents in microblog streams and proposes a novel generative model called the Bilateral Correspondence Latent Dirichlet Allocation (BC-LDA) model, which can assign flexible associations between pictures and words and is able to allow picture-word co-occurrence with bilateral directions, but also single modal association.
- Channel-adaptive unequal error protection for scalable video transmission over wireless channel28
In the proposed channel-adaptive UEP scheme, a bit allocation algorithm is presented to periodically allocate the available bits among different video layers based on varying channel conditions so as to minimize the end-to-end distortion.
- 27
- Revisiting Adversarial Attacks on Graph Neural Networks for Graph Classification27
This work presents a novel and general framework CAMA to generate adversarial examples via manipulating graph structure and node features, which makes use of Graph Class Activation Mapping and its variant to produce node-level importance corresponding to the graph classification task.
- Adversarially Robust Neural Architecture Search for Graph Neural Networks27
G-RNA is proposed, which designs a robust search space for the message-passing mechanism by adding graph structure mask operations into the search space, which comprises various defensive operation candidates and allows us to search for defensive GNNs.
- Adversarial Attack Framework on Graph Embedding Models with Limited Knowledge27
The theoretical connection between graph signal processing and graph embedding models is investigated, and theGraph embedding model is formulated as a general graph signal process with a corresponding graph filter to design a generalized adversarial attack framework: GF-Attack.
- A Comprehensive Survey on Curriculum Learning27
This article summarizes existing CL designs based on the general framework of Difficulty Measurer + Training Scheduler and further categorize the methodologies for automatic CL into four groups, i.e., Self-paced Learning, Transfer Teacher, RL Teacher, and Other Automatic CL.
- Post-training Quantization for Text-to-Image Diffusion Models with Progressive Calibration and Activation Relaxing26
A novel post-training quantization method PCR (Progressive Calibration and Relaxing) for text-to-image diffusion models, which consists of a progressive calibration strategy that considers the accumulated quantization error across timesteps, and an activation relaxing strategy that improves the performance with negligible cost.
- AutoIAS26
This paper presents Automatic Integrated Architecture Searcher (AutoIAS), a framework that provides a practical and general method to find optimal CTR prediction model architecture in an automatic manner and designs an architecture generator network that explicitly models the architecture dependencies among components and generates conditioned architectures distribution for each component.
- Deep Learning for Learning Graph Representations26
This chapter introduces some basic ideas in graph representation/network embedding as well as some representative models in this chapter and explains the motivation for graph representation.
- Non-transitive Hashing with Latent Similarity Components26
The proposed MuCH method significantly outperforms the state-of-art hashing methods especially on search efficiency and proposes a similarity measure, called Multi-Component Similarity, aggregating Hamming similarities in multiple hash tables to capture the non-transitive property of semantic similarity.
- 26
- 26
- 25
- Perceptual Visual Reasoning with Knowledge Propagation25
The proposed Perceptual Visual Reasoning (PVR) model is a module-based approach which incorporates the concept of logical and/or for logic inference, introduces a richer group of perceptual modules for better logic generalization and utilizes the supervised information on each sub-module for more explainability.
- 25
- A Cross-Layer QoS-Support Framework for Multimedia Delivery over Wireless Internet25
This paper presents a framework, which provides QoS (Quality of Service) support, for multimedia delivery over wireless Internet with cross-layer architecture combining the application-level, transport-layer, and link-layer controls.
- A Cross-layer QoS-Supporting Framework for Multimedia Delivery over Wireless Internet25
This paper presents a framework, which provides QoS (Quality of Service) support, for multimedia delivery over wireless Internet with cross-layer architecture combining the application-level, transport-layer, and link-layer controls.
- Multimodal Graph Neural Architecture Search under Distribution Shifts24
A novel Out-of-distribution Generalized Multimodal Graph Neural Architecture Search (OMG-NAS) method which optimizes the MGNN architecture with respect to its performance on decorrelated OOD data and proposes a multimodal graph representation decorrelation strategy.
- Controllable Video Captioning with an Exemplar Sentence24
The proposed Syntax Modulated Caption Generator (SMCG) takes video semantic representation as an input, and conditionally modulates the gates and cells of long short-term memory network with respect to the encoded syntactic information of the given exemplar sentence.
- Come-and-Go Patterns of Group Evolution24
It is found that the evolution patterns of real social groups goes far beyond the classic dynamic models like SI and SIR, and a new model comeNgo is proposed, a concise yet flexible dynamic model for group evolution.
- 24
- 24
- Pasa: power adaptation for starvation avoidance to deliver wireless multimedia24
A distributed dynamic power control scheme termed "power adaptation for starvation avoidance" (PASA), which dynamically adjusts the transmission power of a node so as to avoid starvation, is proposed and shown to be effective in breaking channel captures, hence improving short-term fairness among contending flows.
- Data-Augmented Curriculum Graph Neural Architecture Search under Distribution Shifts23
Data-augmented Curriculum Graph Neural Architecture Search (DCGAS) is proposed, which learns an architecture customizer with good generalizability to data under distribution shifts and outperforms existing methods.
- 23
- A Closer Look at Debiased Temporal Sentence Grounding in Videos: Dataset, Metric, and Approach23
A novel causality-based Multi-branch Deconfounding Debiasing (MDD) framework for unbiased moment prediction is proposed, and a multi-br branch deconfounder is designed to eliminate the effects caused by multiple confounders with causal intervention.
- 23
- A Closer Look at Temporal Sentence Grounding in Videos: Datasets and Metrics23
This paper proposes to re-organize two widely-used TSGV datasets, and deliberatelyange the moment annotation of the test split to make it different from the training split, and introduces a new evaluation metric "dR@n, IoU@$m$" to calibrate the basic IoU scores.
- Social Embedding Image Distance Learning23
The experimental results show that the proposed SIDL approach makes a marked improvement compared to the state-of-the-art image distance metrics, and an interesting observation is given to shows that the learned image distance can better reflect human cognition.
- Automated Disentangled Sequential Recommendation with Large Language Models22
An automated disentangled sequential recommendation (AutoDisenSeq) model that employs neural architecture search (NAS) and design a search space tailored for automated attention representation in attentive intention-disentangled sequential recommendation with an expressive and efficient space complexity of \(O(n^{2})\) given \(n\) as the number of layers is proposed.
- 22
- Quantifying Structural Patterns of Information Cascades22
This paper explores a large-scale dataset including 432 million information cascades with explicit records of spreading traces, and proposes seven-dimensional metrics, which reflect size and spreading orientation aspects, to quantify the structural characteristics of millions of information cascading.
- 22
- 22
- 22
- 21
- Propagation-based social-aware multimedia content distribution21
This article conducts large-scale measurement of a real-world online social network system to study the social content propagation and proposes a propagation-based social-aware delivery framework using a hybrid edge-cloud and peer-assisted architecture.
- 21
- 21
- 21
- Grounding-Prompter: Prompting LLM with Multimodal Information for Temporal Sentence Grounding in Long Videos20
This work proposes a Grounding-Prompter method, capable of conducting TSG in long videos through prompting LLM with multimodal information, and introduces one-shot In-Context-Learning to boost reasoning through imitation, enhancing LLM in TSG task understanding.
- 20
- 20
- Social Visual Image Ranking for Web Image Search20
This paper proposes a novel image ranking approach to web image search, in which social data from social media platform jointly with visual data is used to improve the relevance between returned images and user intentions (i.e., social relevance).
- 20
- 20
- Throughput and QoS Optimization in IEEE 802.11 WLAN20
This paper proposes a class-based analytical model to support multiple traffic classes in MAC layer and presents an optimal solution which maximizes the throughput while satisfying the QoS requirements.
- Channel-adaptive error control for scalable video over wireless channel20
The present invention relates to novel 3,4-dimethyl-2-hydroxy-5-oxo-2,5-dihydropyrrole compounds substituted on the nitrogen atom to their use in agriculture for combating animal and plant pests and also for regulating plant growth.
- VideoDreamer: Customized Multi-Subject Text-to-Video Generation With Disen-Mix Finetuning on Language-Video Foundation Models19
A novel VideoDreamer framework is proposed, which can generate temporally consistent text-guided videos that faithfully preserve the visual features of the given multiple subjects, and a disentangled motion customization strategy to finetune the temporal modules so that it can generate videos with both customized subjects and motions.
- Retraining-free Model Quantization via One-Shot Weight-Coupling Learning19
A bit-width scheduler is designed to dy-namically freeze the most turbulent bit-width of layers during training, to ensure the rest bit-widths converged prop-erly, and an information distortion mitigation technique is presented to align the behaviour of the bad-performing bit-widths to the well-performing ones.
- 19
- 19
- Facial Image-to-Video Translation by a Hidden Affine Transformation19
This work derives an inverse formulation to the affine transformation, enabling automatic inference of the facial expression intensities from videos, and could improve users' communication experience by enabling them to conveniently and creatively produce expression GIFs, which are popular art forms in online messaging and social networks.
- Fates of Microscopic Social Ecosystems19
This paper analyses the fates of different microscopic social ecosystems, namely the final states of their collective activity dynamics in a real-world online social media with detailed individual level records for the first time.
- Learning-to-Ask19
The Learning-to-Ask framework is proposed, within which the agent learns smart questioning strategies for information seeking and knowledge acquisition by means of deep reinforcement learning and generalized matrix factorization respectively, and a Bayesian approach to represent knowledge is adopted to ensure robustness to noisy user responses.
- Joint user-interest and social-influence emotion prediction for individuals19
A probabilistic graphical model is proposed to combine user-interest in multimodal media content and social-influence in social relations together, where a set of parameters are used to measure their importance in influencing the users' emotions, and they are learnt from the user's historical behaviors.
- Social-oriented visual image search19
This paper proposes a community-specific Social-Visual Ranking (SVR) algorithm to rerank the Web images returned by current image search engines, and demonstrates the importance of both visual factors and social factors, and the advantages of social-visual ranking algorithm for Web image search.
- 19
- 19
- Network-Adaptive Rate Control and Unequal Loss Protection with TCP-Friendly Protocol for Scalable Video Over Internet19
A new network-adaptive rate control and Unequal Loss Protection (ULP) scheme in conjunction with TCP-friendly congestion control for scalable video streaming and Rate-Distortion based bit allocation is proposed to minimize the expected end-to-end distortion.
- 19
- Mixup-Augmented Temporally Debiased Video Grounding with Content-Location Disentanglement18
This paper proposes a Disentangled Feature Mixup (DFM) framework for debiased VG, which is capable of performing unbiased grounding to tackle the temporal bias issue and designs a content-location disentanglement module to remove the spurious effect of temporal biases on video representation.
- 18
- 18
- 18
- Online Compact Convexified Factorization Machine18
An Online Compact Convex Factorization Machine (OCCFM) algorithm that eschews the projection operation with efficient linear optimization steps is proposed and it is proved that the developed algorithm achieves a sub-linear regret bound.
- 18
- A Multiscale Survival Process for Modeling Human Activity Patterns18
A novel approach that models the intensity rate of an individual triggering an activity is proposed and it is demonstrated that the model precisely captures corresponding human dynamics across multiple timescales over five orders of magnitudes.
- Learning Socially Embedded Visual Representation from Scratch18
A novel Socially Embedded VIsual Representation Learning (SEVIR) approach, where an Asymmetric Multi-task CNN (amtCNN) model is proposed to embed user intention learning task into semantic learning task to address the sparsity and unreliability problems in social behavioral data.
- 18
- Disentangled Representation Learning for Multimedia17
This tutorial comprehensively presents DRL from various aspects including motivations, definitions, methodologies, evaluations, applications and model designs for multimedia, pointing out challenges in DRL as well as potential research directions deserving future investigations.
- Out-of-Distribution Generalized Dynamic Graph Neural Network for Human Albumin Prediction17
This paper proposes a framework named Out-of-Distribution Generalized Dynamic Graph Neural Network for Human Albumin Prediction (DyG-HAP), which is able to provide accurate albumin predictions for Intensity Care Unit (ICU) patients during hospitalization and proposes a disentangled dynamic graph attention mechanism to capture and disentangle patterns.
- Causal Discovery in Temporal Domain from Interventional Data17
This paper introduces TECDI, a novel approach for temporal causal discovery that leverages the smooth, algebraic characterization of acyclicity in causal graphs to efficiently uncover causal relationships.
- Dynamic Spatio-Temporal Modular Network for Video Question Answering17
A Dynamic Spatio-Temporal Modular Network (DSTN) model is proposed, which utilizes a spatio-temporal modular network to simulate the compositional reasoning procedure of human beings.
- InvisibleFL: Federated Learning over Non-Informative Intermediate Updates against Multimedia Privacy Leakages17
A privacy-preserving solution that avoids multimedia privacy leakages in federated learning by devise a novel encryption scheme called Non-Informative Transformation (NIT) for federated aggregation to eliminates residual multimedia features in intermediate updates and propose Just-Learn-over-Ciphertext (JLoC) mechanism for federate learning.
- Long Short Memory Process17
This work examines the detailed growth process of "WeChat", the largest online social network in China, with 300 million users and 4.75 billion links spanning two years, and proposes three key ingredients, namely average-effect, multiscale-effect and correlation-effect which govern the observed growth patterns at microscopic level.
- 17
- Analyzing social media via event facets17
A rich-media analysis system to address the important challenge of sensing and exploring events from social media in real-time, which includes a novel bilateral correspondence topic model to extract representative content and meaningful facets about events over time.
- 17
- 17
- Automated Graph Machine Learning16
This work extensively discusses automated graph machine learning approaches, covering hyper-parameter optimization (HPO) and neural architecture search (NAS) for graph machine learning, and briefly overview existing libraries designed for either graph machine learning or automated machine learning, respectively.
- PRANCE: Joint Token-Optimization and Structural Channel-Pruning for Adaptive ViT Inference16
PRANCE is introduced, a framework which can jointly optimize activated channels and tokens on a per-sample basis, aiming to accelerate ViTs’ inference process from a unified data and architectural perspective, and introduces a novel “Result-to-Go” training mechanism that models ViTs’ inference process as a Markov decision process, significantly reducing action space and mitigating delayed-reward issues during training.
- 16
- Out-of-Distribution Generalized Dynamic Graph Neural Network with Disentangled Intervention and Invariance Promotion16
This work proposes Disentangled Intervention-based Dynamic graph Attention networks with Invariance Promotion (I-DIDA) to handle spatio-temporal distribution shifts in dynamic graphs by discovering and utilizing invariant patterns, i.e., structures and features whose predictive abilities are stable across distribution shifts.
- 16
- Spectral Graph Attention Network16
Spectral Graph Attention Network (SpGAT) is presented that learn representations for different frequency components regarding weighted filters and graph wavelets bases and can better capture global patterns of graphs in an efficient manner with much fewer learned parameters than that of GAT.
- 16
- 16
- 16
- Causal-aware Graph Neural Architecture Search under Distribution Shifts15
To handle the distribution shifts in NAS process by exploiting the causal graph-architecture relationship to search for optimal architectures that can generalize under distribution shifts, a novel approach, Causal-aware Graph Neural Architecture Search (CARNAS), is proposed.
- SEAM: Searching Transferable Mixed-Precision Quantization Policy through Large Margin Regularization15
A novel method for efficiently searching for effective MPQ policies using a small proxy dataset instead of the large-scale dataset used for training the model, offering several advantages, including high proxy data utilization, no excessive hyper-parameter tuning, and high searching efficiency.
- 15
- Long-term multivariate time series forecasting in data centers based on multi-factor separation evolutionary spatial–temporal graph neural networks15
Experimental results on datasets comprising of temperature and IT power data collected from real-world data centers show that the proposed method outperforms other advanced prediction methods in terms of prediction accuracy, and the learned latent graphs are explainable.
- CurML: A Curriculum Machine Learning Library15
CurML is developed, the first Curriculum Machine L earning library to integrate existing CL algorithms into a unified framework, and is convenient to use and flexible to customize by calling the provided five APIs.
- 15
- Semi-supervised Deep Quantization for Cross-modal Search15
A novel semi-supervised deep quantization (SSDQ) model is proposed that takes both supervised and unsupervised information into account and is capable of incorporating the above three kinds of information into one single framework when utilizing composite quantization for accurate and efficient queries across different domains.
- Understanding Gaming Experience in Mobile Multiplayer Online Battle Arena Games15
This paper is the first to conduct large-scale measurement to study gaming experience of OMGs in the wild, and provides insights that can enhance OMG's system from the perspective of resource allocation and real-time gaming experience monitoring.
- Multi-Modal Generative AI: Multi-Modal LLMs, Diffusions, and the Unification14
A comprehensive overview of multi-modal generative AI, including multi-modal LLMs, diffusions, and the unification for understanding and generation, including key designs including autoregressive-based and diffusion-based modeling, as well as dense and Mixture-of-Experts (MoE) architectures.
- Curriculum Learning: Theories, Approaches, Applications, Tools, and Future Directions in the Era of Large Language Models14
This tutorial focuses on curriculum learning (CL), an important topic in machine learning, which gains an increasing amount of attention in the research community, and advocates novel, high-quality approaches, as well as innovative solutions to the challenging problems in CL.
- Cross-domain Recommendation with Behavioral Importance Perception14
A generic behavioral importance-aware optimization framework for cross-domain recommendation (BIAO) is proposed which predicts the importance of each source behavior according to the corresponding item’s global impact and local user-specific impact and consistently improves the performance of different cross- domain recommendation models.
- Continual Few-shot Learning with Transformer Adaptation and Knowledge Regularization14
A Continual few-shot learning algorithm with Semantic knowledge Regularization (CoSR) for adapting to the distribution changes of visual prototypes through a Transformer-based prototype adaptation mechanism, which improves the model’s ability to distinguish between new and old categories, thus significantly mitigating the catastrophic forgetting problem in continual few- shot learning.
- Curriculum-NAS: Curriculum Weight-Sharing Neural Architecture Search14
Curriculum-NAS is proposed, a curriculum training framework on weight-sharing NAS, which dynamically changes the training data weights during the searching process, so that the potentially optimal architectures can obtain higher probability of being fully trained and discovered.
- 14
- 14
- 14
- Emotionally Representative Image Discovery for Social Events14
This paper builds a hybrid link graph for images of each social event, where the weight of each link is measured by textual emotion information, visual similarity and social similarity, and proposes the Visual-Social-Textual Rank (VSTRank) algorithm to calculate the importance score for each image.
- 14
- 14
- 14
- 14
- 13
- 13
- Adaptive Compression for Online Computer Vision: An Edge Reinforcement Learning Approach13
A reinforcement learning (RL)-based adaptive JPEG configuration framework, AdaCompress, which designs an edge RL agent that learns the optimal compression quality level to achieve an expected inference accuracy and upload image size, only from the online inference results, without knowing details of the model structures.
- 13
- AdaCompress13
An agent is designed that adaptively chooses the compression level according to the input image's features and backend deep learning models and this approach can reduce the size of images by 1/2 -- 1/3 while the overall classification accuracy only decreases slightly.
- 13
- ACM TIST Special Issue on Data-Driven Intelligence for Wireless Networking13
The aim of this special issue is to promote the state of the art in scientific and practical research on big data intelligence for wireless network design and to bring together researchers and practitioners from academia and industry in an effort to present their research work and share research and development ideas.
- Dispersing Instant Social Video Service Across Multiple Clouds13
This paper analyzes and formulates the multi-cloud hosting of an instant social video system as an optimization problem, and proposes a heuristic algorithm with acceptable complexity to solve the optimization problem.
- 13
- 13
- Distributed distance measurement for large-scale networks13
Group-based distance measurement service (GDMS), which estimates and disseminates distance information of node-pairs in large-scale wide area networks and a novel multicast-based algorithm is used for both intra- and inter-group performance measurements.
- 13
- 13
- Large Language Model with Curriculum Reasoning for Visual Concept Recognition12
A novel rule enhancement module with a tool library, which fully leverage the reasoning capability of large language models and can generate human-understandable rules without any annotation, and a curriculum data resampling methodology to help the large language model accurately extract from easy to complex rules at different reasoning stages.
- LLM4VG: Large Language Models Evaluation for Video Grounding12
The LLM4VG benchmark is proposed, which systematically evaluates the performance of different LLMs on video grounding tasks and proposes prompt methods to integrate the instruction of VG and description from different kinds of generators, including caption-based generators for direct visual description and VQA-based generators for information enhancement.
- Probabilistic Attributed Hashing12
This paper proposes a hashing learning framework, Probabilistic Attributed Hashing (PAH), to integrate attributes with low-level features, and develops an efficient iterative learning algorithm, which is generally feasible for large-scale applications.
- 12
- Advances in wireless video12
This special issue of IEEE Wireless Communications reflects their state of the art, including recent advances in wireless video coding and delivery, as well as supporting wireless network technologies.
- Towards Lightweight Graph Neural Network Search with Curriculum Graph Sparsification11
The paper proposes to design a joint graph data and architecture mechanism, which identifies important sub-architectures via the valuable graph data, which achieves on-par or even higher node classification performance with half or fewer model parameters of searched GNNs and a sparser graph.
- 11
- Continual Recognition with Adaptive Memory Update11
This article proposes a continual recognition model with adaptive memory update that is capable of overcoming the problem of catastrophic forgetting with various new classes coming in sequence, especially for similar but different classes.
- Module-Aware Optimization for Auxiliary Learning11
By approximating all previous θτ as θM, the derivation of ∇αθM requires the Jacobi and Hessian matrix in previous M steps, which is memory consuming.
- Automated Machine Learning on Graph11
This tutorial is the first to systematically and comprehensively review automated machine learning on graphs, possessing a great potential to draw a large amount of interests in the community.
- 11
- 11
- 11
- 11
- Uncovering the Co-driven Mechanism of Social and Content Links in User Churn Phenomena11
A novel survival model is proposed, which incorporates both social and content factors, to predict churn probability over time and successfully fits the churn distribution in reality and accurately predicts the churn rate of different subpopulations in the future.
- The General Black-box Attack Method for Graph Neural Networks.11
This paper begins by investigating the theoretical connections between different kinds of GNNs in a principled way and integrate different GNN models into a unified framework, dubbed as General Spectral Graph Convolution, and proposes a generalized adversarial attacker, which does not require any knowledge of the target classifiers used inGNNs.
- Wireless Caching in Large-Scale Edge Access Points11
The NP-hardness of the wireless caching problem is proved, an objective function that maximizes the total cache hit rate of all APs is introduced and a local distributed caching algorithm is proposed to address it.
- 11
- Multi-modal Sequence to Sequence Learning with Content Attention for Hotspot Traffic Speed Prediction11
This work proposes Multi-Seq2Seq-Att, a multi-modal sequence learning model that deals with two sequences in different modalities, namely, the query sequence and the traffic speed sequence, and addresses the modality gap between queries and the Traffic speed.
- 11
- 11
- Fast Segmentation of Abdominal Wall: Application to Sliding Effect Removal for Non-rigid Registration11
This paper shows why and how a quick interactive delineation of the full abdominal wall (AW) can be performed using B-spline interpolation and demonstrates the approach is a promising method to remove the sliding motion effect on the non-rigid registration of abdominal images.
- 11
- 11
- 11
- 11
- 11
- 11
- 11
- TMPQ-DM: Joint Timestep Reduction and Quantization Precision Selection for Efficient Diffusion Models10
TMPQ-DM is introduced, which jointly optimizes timestep reduction and quantization to achieve a superior performance-efficiency trade-off, addressing both temporal and model optimization aspects.
- 10
- Preserving Ordinal Consensus: Towards Feature Selection for Unlabeled Data10
An unsupervised triplet-induced graph is proposed to explore a new type of potential structure at feature level, and incorporates it into simultaneous feature selection and clustering, and an alternating minimization algorithm is developed to efficiently optimize the proposed model.
- Exploring the collective human behavior in cascading systems: a comprehensive framework10
This paper examines a real-world online social media with more than 1.7 million information spreading records and proposes a comprehensive generative framework with a latent user interest layer to capture the collective behavior and accurately models the information cascades.
- Affective Contextual Mobile Recommender System10
This paper proposes Kaleido, a real mobile system to achieve an affect-aware learning-based social media recommendation that achieves speed-up of a factor of 1,000 against the local data training execution on smartphones.
- 10
- 10
- 10
- Modular Machine Learning: An Indispensable Path towards New-Generation Large Language Models9
The integration of the MML with LLMs has the potential to bridge the gap between statistical (deep) learning and formal (logical) reasoning, thereby paving the way for robust, adaptable, and trustworthy AI systems across a wide range of real-world applications.
- Self-directed machine learning9
SDML is designed as a self-directed learning process guided by self-awareness, including internal awareness and external awareness that could enable machines to conduct human-like self- directed learning and provide a new perspective towards artificial general intelligence.
- 9
- 9
- 9
- MPEG-4 video transport over the Internet: a summary9
The proposed MPEG-4 video transport architecture ties together a range of technologies from both signal processing and networking areas, and provides important experience and insights on how to harness the power of these technologies in designing protocols and mechanisms for MPEG- 4 video transport over the Internet.
- 9
- 9
- User-aware object-based video transmission over the next generation Internet9
A new transport framework for complex multimedia applications over the next generation Internet, which provides differentiation functionality within one IP session as well as among different IP sessions, and a new bitstream classification, prioritization and packetization scheme.
- Regularized total least squares approach for nonconvolutional linear inverse problems9
In this correspondence, a solution is developed for the regularized total least squares estimate in linear inverse problems where the linear operator is nonconvolutional by modifying the RQ function to enforce a smooth solution.
- Use of second-order derivative-based smoothness measure for error concealment in transform-based codecs9
Simulation results show that a weighted combination of the quadratic variation and the Laplacian operator can significantly reduce the blurring across the edges while enforcing smoothness along the edges.
- Modular-Cam: Modular Dynamic Camera-view Video Generation with LLM8
To better understand a given complex prompt, a large language model is utilized to analyze user instructions and decouple them into multiple scenes together with transition actions and proves Modular-Cam's strong capability of generating multi-scene videos together with its ability to achieve fine-grained control of camera movements.
- 8
- 8
- Neighbor Does Matter: Curriculum Global Positive-Negative Sampling for Vision-Language Pre-training8
A curriculum-based Global Positive-Negative Sampling (GPN-S) framework for vision-language pre-training is proposed, which conducts both positive and negative sampling in the global level, grounded on the notion of neighborhood relationships.
- 8
- Meta Hyperparameter Optimization with Adversarial Proxy Subsets Sampling8
A novel meta hyperparameter optimization model with an adversarial proxy subsets sampling strategy (Meta-HPO), which can transfer hyperparameters optimized on the sampled proxy subset to the full dataset and further adapt to the new data in an out-of-sample updating manner, and a meta transfer framework, named "hp-learner'', to build the connection between the distribution of dataset and the optimal hyper parameters.
- Joint User- and Event- Driven Stable Social Event Organization8
This paper proposes an efficient greedy heuristic algorithm to solve Stable-SEO, taking both user preferences and event preferences into account, and is the first to study SEO with both preferences from users to events and preferences from events to users being considered.
- 8
- An efficient transport scheme for multimedia transport over wireless Internet8
A method of manufacturing connection-type ceramic packages for integrated circuits is disclosed, where a plurality of green ceramic sheets are prepared and punched to form notches in each sheet and recesses defining cavities for accommodating an integrated circuit element therein at predetermined positions in given sheets.
- Behavior Importance-Aware Graph Neural Architecture Search for Cross-Domain Recommendation7
This work proposes Behavior importance-aware Graph Neural Architecture Search (BiGNAS), a framework that jointly optimizes GNN architecture and data importance for CDR, and is the first work to jointly optimize GNN architecture and behavior data importance for cross-domain recommendation.
- 7
- ScenarioDiff: Text-to-video Generation with Dynamic Transformations of Scene Conditions7
This work proposes a model named ScenarioDiff, able to generate longer videos with scene transformations, which employs a spatial layout fuser to control the positions of subjects and the scenes of each frame and introduces mixed frequency controlnet.
- Lightweight Diffusion Models with Distillation-Based Block Neural Architecture Search7
This work proposes to automatically remove the structural redundancy in diffusion models with the proposed Diffusion Distillation-based Block-wise Neural Architecture Search (DiffNAS), and proves the necessity of the dynamic adjustment of this loss.
- 7
- 7
- Uncovering Pattern Formation of Information Flow7
By exploring 432 million information flow patterns extracted from a large-scale online social media dataset, a wide range of complex geometric patterns characterized by a three-dimensional metric space are uncovered, potentially leading to wide implications for prediction, control and policy decisions in social media.
- Dynamical Origins of Distribution Functions7
A theorem is proposed which builds a mathematical relationship between a dynamical system modeled by differential equations and the distribution function (or survival function) of the cross-sectional states of this system and this framework is able to capture the continuous-time dynamics of an evolutionary system.
- Learning and Interpreting Complex Distributions in Empirical Data7
This paper showcases a four-parameter dynamic model together with inference and simulation algorithms, which is able to fit and generate a family of distributions, ranging from Gaussian, Exponential, Power Law, Stretched Exponential (Weibull), to their complex variants with multi-scale complexities.
- Social Media Profiler: Inferring Your Social Media Personality from Visual Attributes in Portrait7
Comprehensive experiments demonstrate that the proposed method to infer social media personality from portrait can achieve significant performance gain over the existing method.
- Image Segmentation Based on Improved Otsu Algorithm7
The results indicate that the improved Otsu method characterized by the combination of the region information and edge information can lead to more accurate and effective segmentation.
- 7
- 7
- 7
- 7
- A fast and memory efficient algorithm for down-conversion of an HDTV bitstream to an SDTV signal7
This paper proposes a fast and memory efficient algorithm for the down-conversion of an HDTV bitstream down to an SDTV signal by combining inverse discrete cosine transform (IDCT) with the pixel-domain down-Conversion.
- 7
- U2USim - A UAV Telepresence Simulation Platform with Multi-agent Sensing and Dynamic Environment6
U2USim is introduced, a telepresence simulation platform with a dynamic environment, serving as a realistic synthetic data generation, performance evaluation, and visualization tool for UAV-to-UAV (U2U) cooperative learning.
- 6
- 6
- Watch, Reason and Code6
A Watch-Reason-Code (WRC) model to synthesise programs that describe the process carried out in a set of video sequences and can achieve the state-of-the-art under a variety of settings is proposed.
- Sketched Follow-The-Regularized-Leader for Online Factorization Machine6
This paper proposes Generalized Frequent Directions to approximate indefinite symmetric matrices in a streaming way, making that the sum of historical gradients for FM could be estimated with tighter error bound efficiently.
- come N go6
A new model comeNgo is proposed, a concise yet flexible dynamic model for group evolution that generalizes earlier theoretical models and different joining and quitting mechanisms the authors find from observation, and can provide insights about different evolution patterns of social groups.
- 6
- 6
- 6
- Network-Adaptive Cache Management Schemes for Mixed Media6
A media-characteristic-weighted replacement policy is presented to improve the hit ratio of mixed media including continuous and noncontinuous media, and a network-condition- and media-quality-adaptive resource management mechanism is described to dynamically re-allocate cache resource for different types of media according to their request patterns.
- User- and content-aware object-based video streaming over the Internet6
Experimental results show that the proposed scheme provides must improved QoS and capability for object-based user interactivity.
- Aligning Large Multimodal Model with Sequential Recommendation via Content-Behavior Guidance5
AlmostRec is a novel framework that incorporates multimodal information, including historical interaction IDs, textual descriptions, and images of items into large foundation models, to facilitate controllable predictions in multimodal scenarios.
- Towards multimodal graph large language model5
This work explores the potential of multimodal graph large language models (MG-LLM) to unify and generalize across diverse multimodal graph data and tasks, proposing a unified framework of multimodal graph data, tasks, and models.
- 5
- Global-Local GraphFormer: Towards Better Understanding of User Intentions in Sequential Recommendation5
The global graph information is utilized to help the Transformer-based sequential recommendation, where the information from different modalities, i.e., user-item interactions in the global graph and the temporal patterns in the historical sequences are taken into account jointly.
- 5
- Out-of-distribution Generalization and Its Applications for Multimedia5
This tutorial is to disseminate and promote the recent research achievements on out-of-distribution generalization as well as their applications on multimedia, which is an exciting and fast-growing research direction in the general field of machine learning and multimedia.
- 5
- 5
- Learning From Networks5
This tutorial systematically review the area of learning from networks, including algorithms, theoretical analysis, and illustrative applications, and covers the latest hot topic, graph neural based approaches.
- 5
- 5
- Guest Editorial Video Distribution Over Future Internet5
The current Internet is under tremendous pressure due to the exponential growth in bandwidth demand, fueled by the transfer of video consumption to online distribution, IPTV, streaming services such as Netflix, and from phone networks to videoconferencing and Skype-like video communications.
- 5
- Low-delay cloud-based rendering of free viewpoint video for mobile devices5
The proposed low-delay cloud-based FVV rendering framework can improve the quality of experience substantially with improved free viewpoint video quality and low interaction delay on mobile devices.
- 5
- Adaptive QoS control for real-time video communication over wireless channels5
This paper proposes adaptive quality-of-service (QoS) control for real-time video communication over wireless channels that is capable of achieving bounded delay, high reliability, and efficiency.
- 5
- 5
- 5
- 5
- 4
- Curriculum Learning for Multimedia in the Era of Large Language Models4
This tutorial focuses on curriculum learning (CL), an important topic in machine learning, which gains an increasing amount of attention in the research community and is attractive to the audience interested in machine learning and multimedia from both academia and industry.
- 4
- Multimedia Cognition and Evaluation in Open Environments4
What and how efforts have been done in the literature are discussed and insights on research directions that deserve further study are shared to produce potentially profound impacts on multimedia cognition and evaluation in open environments are shared.
- Uncovering the Heterogeneous Effects of Preference Diversity on User Activeness: A Dynamic Mixture Model4
This paper examines large-scale real-world datasets collected from two of the most popular video-sharing social platforms in China, including the behavior logs of more than 787 thousand users and 1.95 million videos, with accurate content category information, and proposes a universal mixture model with the capability of accurately fitting dynamic activeness curves while reflecting the heterogeneous patterns of preference diversity.
- Permutation-equivariant and Proximity-aware Graph Neural Networks with Stochastic Message Passing4
This paper proposes Stochastic Message Passing (SMP) model, a general and simple GNN to maintain both proximity-awareness and permutation-equivariance and theoretically proves that the mechanism can enable GNNs to preserve node proximities, and at the same time, maintain permutations-Equivariance with certain parametrization.
- 4
- 4
- 4
- 4
- AutoML and Meta-learning for Multimedia4
This tutorial will first advocate novel, high-quality research findings and innovative solutions to the challenging problems in Auto ML and meta-learning, and point out future research directions on AutoML andmeta-learning as well as their potential new applications for multimedia.
- Social Video Content Delivery4
This brief presents new architecture and strategies for distribution of social video content, identifying the unique characteristics of socially-aware video access and social content propagation, and revealing the design and integration of individual modules that are aimed at enhancing user experience in the social network context.
- 4
- 4
- 4
- End-to-End Mobility Support in IPv6 Using Peer-to-Peer Technologies4
The experiments with the testbed demonstrated that a complete end-system based mobility solution is technically feasible and should be easy to deploy in the real world without the need of introducing new network components.
- 4
- 4
- 4
- 4
- Disentangled Graph LLM for Molecule Graph Editing under Distribution Shifts3
The proposed MoFE model is a disentangled graph large language model for molecule graph editing that handles editing instructions under distribution shifts via disentangling invariant factors that govern editing-relevant properties, and promotes joint disentanglement between experts and latent factors, reinforcing their alignment and preventing collapse.
- Out-of-Distribution Generalized Graph Anomaly Detection with Homophily-aware Environment Mixup3
This study proposes Ego-Neighborhood Disentangled Encoder with H omophily-aware E nvironment M ixup ( HEM), which effectively handles structural distribution shifts in GAD by discovering invariant patterns.
- Quantization Meets OOD: Generalizable Quantization-aware Training from a Flatness Perspective3
This paper proposes a flatness-oriented QAT method, FQAT, which introduces a layer-wise freezing mechanism to mitigate the gradient conflict issue between dual optimization objectives and proposes an disorder-guided adaptive freezing algorithm, effectively addressing the challenges caused by interference between layers.
- Identity-Text Video Corpus Grounding3
Experimental results demonstrate the effectiveness of the proposed Video-Locator model and highlight the importance of identity-generalization capability for ITVCG.
- Revisiting Transformation Invariant Geometric Deep Learning: An Initial Representation Perspective3
It is proved that TinvNet can strictly guarantee transformation invariance, being general and flexible enough to be combined with the existing neural networks, and advocated that TinvNet should be considered as an essential baseline for further studies of transformation-invariant geometric deep learning.
- Decoding cell identity with multi-scale explainable deep learning3
The proposed Cell Decoder is a biological prior knowledge informed model, offering multi-view interpretability for decoding cell identity and data integration, and has shown its applicability in uncovering novel cell types and states in both human bone and mouse embryonic contexts.
- 3
- 3
- Domain Generalization through the Lens of Angular Invariance3
A novel angular invariance and the accompanied norm shift assumption is put forward and a novel deep DG method dubbed Angular Invariance Domain Generalization Network (AIDGN) is proposed, developed with a von-Mises Fisher mixture model.
- FastPR: One-stage Semantic Person Retrieval via Self-supervised Learning3
This work proposes FastPR, a one-stage semantic person retrieval method via self-supervised learning, to optimize the person localization and semantic retrieval simultaneously, and proposes a dynamic visual-semantic alignment mechanism which utilizes grid-based attention to fuse the cross-modal features, and employs a label prediction proxy task to constrain the attention process.
- A Simple and General Graph Neural Network with Stochastic Message Passing3
Stochastic Message Passing (SMP) is proposed, a general and simple GNN to maintain both proximity-awareness and permutation-equivariance properties and it is proved that such a mechanism can enable GNNs to preserve node proximities in theory while maintaining permutations-Equivariance with certain parametrization.
- 3
- Guest Editorial Multimedia Economics for Future Networks: Theory, Methods, and Applications3
Multimedia data is thus becoming the dominant traffic in the near future for both wired and wireless networks.
- 3
- Rethinking HTTP Adaptive Streaming with the Mobile User Perception3
This paper identifies a novel research problem on excessive visual quality which leads to bitrate-inefficient video streaming, and proposes a flexible system called EyeTube to address it, and applies dynamic resolution scaling on mobile video streaming to trade off the bitrate efficiency and user viewing experience.
- Optimal Resource Allocation for Distributed Video Communication3
Comprehensive and systematic, Optimal Resource Allocation for Distributed Video Communication presents a unified optimization framework for resource allocation across these systems, providing the required foundation in video communications.
- 3
- mPano: cloud-based mobile panorama view from single picture3
This paper presents a novel cloud-based mobile panorama view system that can generate and view panorama-view on mobile devices from a single picture, namely mPano and proposes a novel iterative multi-modal image retrieval (IMIR) approach to get spatially adjacent images using both tag and content information from the single picture.
- Cloud Computing and Dynamic Resource Allocation for Multimedia Applications3
This special issue contains five papers, in which two papers are related to cloud-based multimedia applications, two papers deal with Peer-to-Peer (P2P) video streaming systems, and one paper addresses the resource allocation problem in wireless networks.
- A Novel Energy Saving Approach through Mobile Collaborative Computing Systems3
This paper proposes a novel energy savings approach through mobile collaborative systems, which jointly consider computing and communication energy cost and builds a mathematical energy model for collaborative computing systems.
- 3
- 3
- 3
- 3
- 3
- Guest Editorial Advanced Mobility Management and QoS Protocols for Next-Generation Wireless Internet3
- 3
- Wavelet Based Multigrid Reconstruction Algorithm for Optical Tomography3
The experimental results show that the multigrid method can achieve similar reconstruction quality with a significantly reduced amount of time than the one-grid algorithm.
- Regularized multichannel restoration of color images using cross-validation3
Deterministic multichannel filters are proposed that do not utilize any prior knowledge about the multich channel image and the noise and a technique to estimate the variance of the multICHannel noise is proposed.
- OOD-GraphLLM: Graph Large Language Model for Out-of-Distribution Generalized Drug Synergy Prediction2
OOD-GraphLLM is proposed, a novel graphLLM framework which is able to accurately predict drug synergy under O.O.D.D. settings via jointly optimizing molecular graph representation and biomedical semantic language representations in a unified manner.
- AV-Unified: A Unified Framework for Audio-Visual Scene Understanding2
AV-Unified is proposed, a unified framework that enables joint learning across a wide range of audio-visual scene understanding tasks and incorporates a multi-scale spatiotemporal perception network to effectively capture audio-visual associations.
- Autonomous machine learning2
Inspiration from human learning sets the focus on one essential but poorly studied characteristic of learning: Autonomy, with powerful algorithms surpassing human performances in some very specific domains of expertise, but these models still have very poor autonomy.
- DyNAS-DDI: Dynamic Pairwise Architecture Search for Generalizable Drug-Drug Interaction LLM2
DyNamic Pairwise Architecture Search for Generalizable Drug-Drug Interaction LLM (DyNAS-DDI), a novel framework that dynamically adapts network architectures for each molecular pair and integrates biomedical knowledge from LLMs to improve generalization under distribution shifts is proposed.
- Graph Machine Learning under Distribution Shifts: Adaptation, Generalization and Extension to LLM2
This tutorial is to disseminate and promote the recent research achievement on graph out-of-distribution adaptation, graph out-of-distribution generalization, and large language models for tackling distribution shifts, which are exciting and fast-growing research directions in the general field of machine learning and data mining.
- Improving Compositional Generalization in Cross-Embodiment Learning via Mixture of Disentangled Prototypes2
A novel Mixture of Disentangled Prototypes (MoDP) method to improve the compositional generalization in Cross-Embodiment Learning by introducing a finite prototype space that bridges the gap between unbounded agent-environment combinations and a single policy model.
- GAQAT: gradient-adaptive quantization-aware training for domain generalization2
A novel Gradient-Adaptive Quantization-Aware Training (GAQAT) framework for DG, which introduces a mechanism to quantify gradient inconsistencies and selectively freeze the gradients of scaling factors, thereby stabilizing the training process and enhancing out-of-domain generalization.
- 2
- Imputation with Inter-Series Information from Prototypes for Irregular Sampled Time Series2
This work proposes PRIME, a Prototype Recurrent Imputation ModEl, which integrates both intra-series and inter-series information for imputing missing values in irregularly sampled time series and underscores PRIME's superiority over the state-of-the-art models by up to 26% relative improvement on mean square error.
- Traffic-Driven Proactive Slot Orchestration in an Optical Network with Fine-Granularity Flexible Ethernet Interfaces2
A traffic-driven proactive FlexE slot orchestration algorithm called PMFAN-GDSO is proposed, based on this algorithm, bandwidth variations are calculated in advance based on the bandwidth change provided via the traffic prediction algorithm before service fluctuations occur, avoiding the time-consuming slot Orchestration calculations and service abrupt changes.
- 2
- 2
- 2
- 2
- Inter-and-Intra Domain Attention Relational Inference for Rack Temperature Prediction in Data Center2
Experiments show that I2A-RI outperforms other state-of-the-art models since it takes the advantage of the ability to infer the potential interactions across domains and utilizes the inferred interaction structure to achieve accurate dynamical predictions.
- 2
- Commonsense Learning: An Indispensable Path towards Human-centric Multimedia2
The basic concept of commonsense knowledge and commonsense reasoning is introduced, commonsense resources and benchmarks are summarized, an overview on recent commonsense learning and reasoning methods are given, and several popular applications of commonsens knowledge in real-world scenarios are discussed.
- Hardware-aware One-Shot Neural Architecture Search in Coordinate Ascent Framework2
This paper proposes HURRICANE, a algorithm that explores the automatic hardware-aware search over a much larger search space and a multistep search scheme in coordinate ascent framework, to generate tailored models for different types of hardware.
- 2
- Reading Document and Answering Question via Global Attentional Inference2
A novel global attentional inference (GAI) neural network architecture is proposed, which learns useful cues from structural knowledge via a dynamically terminated multi-hop inference mechanism, to answer cloze-style questions.
- 2
- Researches on multimedia technology 2014—deep learning and multimedia computing2
Familiarity, ease of access, trust, and awareness are key considerations.
- 2
- 2
- 2
- 2
- Multichannel image identification and restoration based on the EM algorithm and cross-validation2
A novel representation of MC signal processing is introduced and this notation leads to a more natural extension of single-channel signal processing algorithms to the MC case and yields a new class of matrices which are called semi-block- circulant (SBC) matrices.
- 1
- Multi-sentence Video Grounding for Long Video Generation1
The approach seamlessly extends the development in image/video editing, video morphing and personalized generation, and video grounding to the long video generation, offering effective solutions for generating long videos at a low memory cost.
- 1
- 1
- 1
- 1
- 1
- H2V4Sports: Real-Time Horizontal-to-Vertical Video Converter for Sports Lives via Fast Object Detection and Tracking1
H2V4Sports is presented, a real-time horizontal-to-vertical video converter specifically designed for sports live broadcasts and proposes a video virtual director platform that captures the most informative vertical zones from horizontal video live frames using various director logic for a smooth frame- to-frame transition.
- 1
- 1
- Regularized Progressive Expansion Algorithm for Recovery of Scattering Media from Time-Resolved Data1
- 1
- 1
- 1
- 1
- 1
- 1
- CFCR: A Convolution and Fusion Model for Cross-platform Recommendation1
The proposed CFCR model utilizes Graph Convolution Networks (GCN) to extract user and item features on graphs from different platforms, and fuses cross-platform information by Multimodal AutoEncoder with common latent user features to ensure high-order connectivity information is preserved to the most extent and domain-invariant user representations are automatically obtained.
- 1
- 1
- Analyzing WeChat Diffusion Cascade: Pattern Discovery and Prediction1
An analysis framework for WeChat cascade based on the characteristics of cross-scenario diffusion is proposed and some typical diffusion patterns are revealed by analyzing a real WeChat dataset.
- 1
- [Study of the Application of Mobile Medical Technology in Construction of Grading Diagnosis and Treatment System].1
Four measures are proposed for further developing mobile medical application in the GDTS: the government departments should formulate policies and industry standards of products as soon as possible to meet requirement of market; service providers should take the hospitals as core role to achieve mutual benefit and win-win situation.
- Online Social Media Content Delivery1
A joint online processing and delivery paradigm to improve the user experience in social video services is proposed to satisfy the trend for content to be dynamically processed before being delivered to users.
- Joint Content- and Social-Based User Preference Mining1
The chapter presents a general framework to understand user preference, which can be inferred from both content and social aspects in the context of online social content delivery.
- Propagation-Based Social Video Content Replication1
This chapter presents propagation-based social-aware delivery studies, which aim to study the service provision of social content for global users that provide a satisfactory quality of experience (QoE) in this chapter.
- From Global to Local: A Context-Embedded LSTM Recurrent Network for Local Content Popularity Prediction1
This paper reveals the characteristics of “local popularity”, and proposes a context-embedded LSTM recurrent network, leveraging the correlation between the geographical context and content popularity to estimate a fine-grained content popularity.
- 1
- Enhancing Multimedia Network Resource Allocation Using Social Prediction1
This chapter presents strategies to enhance the performance of network resource allocation based on the prediction of social behaviors to support resource allocation in an online social network.
- 1
- Segmentation and registration of CT multi-phase images for abdominal surgical planning1
The purpose of this thesis is to provide an accurate registration method for abdominal viscera between venous and arterial phase CT images by separating the image into big motion and less motion regions, and performing the registration on new images where abdo-thoracic wall and thoracic Viscera are removed.
- Real-time 3D Applications on Handheld Devices: Challenges and Trend1
Developing and appreciating 3D graphics and animations has become a requirement not only in research projects, but also in many practical applications and daily activities, such as advertising, movies, games, TV, training and rehabilitation.
- 1
- 1
- 1
- Utilizing the Diversities and Invariants of the Internet to Support Multi-Services1
Simulation results indicate that a simple, scalable, and easy-to-deploy scheme can be designed to support multi-services by utilizing the invariants of the Internet through HARP.
- 1
- 1
- 1
- 1
- 1
- Text-guided Molecule Generation with Conditional Discrete Graph Diffusion Model–
A text-guided molecular graph generation framework that leverages the structural modeling power of graph diffusion models to achieve both strong alignment with textual descriptions and high-quality molecular structures and a molecule structure consistency loss that explicitly enforces structural coherence during generation, leading to higher-quality and more consistent molecular graphs.
- Curriculum-GraphLLM: Joint Optimization of Architectures, Structures and Texts for Denoised Graph Neural Architecture Search–
This paper proposes Curriculum-GraphLLM, a novel graphLLM framework for joint optimization of architectures, structures, and texts for denoised graph neural architecture search to jointly optimize GNN architectures, graph structures, and textual semantics as a unified denoising process during architecture search.
- –
- Continual-GraphLLM: Dynamic Graph Large Language Model with Invariance Regularized Adaptive Multi-Scale Experts–
A novel Continual Learning Dynamic Graph LLM framework (Continual-GraphLLM) is proposed to continually adapt to incoming patterns by routing them to experts specialized in similar past patterns, while mitigating the overwriting of previously learned patterns by assigning new experts to unseen patterns.
- –
- –
- Beyond Graph Distribution Shifts: LLMs, Adaptation, and Generalization–
This tutorial presents a comprehensive overview of three emerging and synergistic directions for tackling distribution shifts in graph learning, which highlight Graph LLMs, which combine the representational power of large language models with graph structures to enable flexible, in-context, and few-shot learning on graphs.
- Reflective Cross-Granularity Grounding with Preference Optimization for Long Video Understanding–
ReCrossVLLM is proposed, a reflective cross-granularity grounding framework for video LLM with preference optimization to collaboratively achieve long video understanding, which not only retains the capabilities of high-level video semantics understanding, but also strengthens the fine-grained understanding abilities.
- –
- Research progress on multimedia intelligence computing Research progress on multimedia intelligence computing–
The interaction mechanisms among media content, user behavior, and network topology are revealed, a media content dissemination prediction model based on user behavior and a large-scale network topology model are proposed, and a new theory for cross-domain collaborative computation of resources for large-scale multimedia data is established.
- –
- –
- –
- Independence-promoted disentangled dynamic graph attention network for out-of-distribution generalization–
A disentangled dynamic graph attention network that explicitly separates invariant patterns from variant ones is introduced and an independence-promoted disentangled disentangled disentangled disentanglement optimization framework is designed to enhance the model’s ability to capture invariant patterns for OOD generalization.
- –
- –
- The Impact of AR Innovation Technology Characteristics on Consumer Satisfaction and Continuous Usage Intention–
It was observed that user satisfaction pertaining to the AR-integrated fashion platforms positively influenced the intent of continuous platform usage, underscoring the significance of understanding AR technology characteristics in enhancing customer satisfaction and encouraging sustained platform usage.
- –
- –
- –
- –
- [Discussion on Technical Evaluation Requirements of Allergen Detection Reagent Pre-marketing].–
This paper discusses and gives regulatory suggestions, with a view to providing technical reference for product R&D, production, evaluation, approval and supervision in this field.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Joint Online Processing and Geo-Distributed Delivery for Dynamic Social Streaming–
This chapter presents some exploration into incorporating content processing in the social content delivery framework.
- –
- T2U–
T2U, a cross-platform video recommendation system with a novel interface, is presented, designed to show not only the recommended videos but also the reasons to users intuitively, thus significantly enhances user experience.
- Human-like Visual Learning and Reasoning–
In this tutorial, several basic problems in human-like visual learning and reasoning will be discussed, including why the authors need to learn from human learning process, what are the research goals of human-like visual learning and reasoning, and how to embed human learning characteristics into learning models.
- Learning to Learn Image Classifiers with Informative Visual Analogy.–
This paper proposes a novel Visual Analogy Network Embedded Regression model, a novel out-of-sample embedding method to learn the embedding of a new class represented by a few samples through its visual analogy with base classes, and investigates the fusion strategy of the two kinds of parameters.
- –
- Popularity of Social Videos–
This work investigates the popularity of videos propagated through online social networks, including the popularity distribution and its evolution, and investigates predictive models that capture social video popularity.
- –
- Dynamical Social Video Propagation–
This chapter takes a closer look at the social propagation process for videos, which determines how individual videos reach different users in online social networks.
- –
- –
- –
- –
- –
- –
- Efficient Mobility Management for Seamless Roaming between WWAN and WLAN–
A novel mobility management system which integrates a connection manager that can intelligently detect the network conditions change and a virtual connectivity manager that uses end-to-end principle to maintain the session connectivity is proposed.
- –
- Maximum entropy restoration method of linearly degraded binary image–
The experimental results show that the maximum entropy restoration method and maximum bounded entropy Restoration method with binary constraint can improve the quality of restored image.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.