Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks437 from public data
- EnlightenGAN: Deep Light Enhancement Without Paired Supervision2,636
This paper proposes a highly effective unsupervised generative adversarial network, dubbed EnlightenGAN, that can be trained without low/normal-light image pairs, yet proves to generalize very well on various real-world test images.
- 857
- 829
- Joint Service Caching and Task Offloading for Mobile Edge Computing in Dense Networks588
An efficient online algorithm, called OREO, is proposed, which jointly optimizes dynamic service caching and task offloading to address a number of key challenges in MEC systems, including service heterogeneity, unknown system dynamics, spatial demand coupling and decentralized coordination.
- Jointly Attentive Spatial-Temporal Pooling Networks for Video-Based Person Re-identification346
This work presents a novel joint Spatial and Temporal Attention Pooling Network (ASTPN) for video-based person re-identification, which enables the feature extractor to be aware of the current input video sequences, in a way that interdependency from the matching items can directly influence the computation of each other's representation.
- Masked Diffusion Transformer is a Strong Image Synthesizer329
A Masked Diffusion Transformer (MDT) is proposed that introduces a mask latent modeling scheme to explicitly enhance the DPMs’ ability to contextual relation learning among object semantic parts in an image.
- 310
- 255
- 239
- Client Selection for Federated Learning With Non-IID Data in Mobile Edge Computing230
This paper proposes an efficient FL algorithm, named CSFedAvg, in which the clients with lower degree of non-IID data will be chosen to train the models with higher frequency, and conducts simulations using publicly-available datasets to train deep neural networks.
- 220
- CLU-CNNs: Object detection for medical images203
A domain adaptation framework called CLU-CNNs is proposed, which is designed for medical images and uses ANCF and BN-IN Net to improve domain adaptation capability without specific domain adaptation training.
- 198
- Context-aware Biaffine Localizing Network for Temporal Sentence Grounding195
A novel localization framework that scores all pairs of start and end indices within the video simultaneously with a biaffine mechanism that significantly outperforms state-of-thearts on three public datasets, demonstrating the effectiveness of the proposed localization framework.
- 190
- 186
- Jointly Cross- and Self-Modal Graph Attention Network for Query-Based Moment Localization152
A novel Cross- and Self-Modal Graph Attention Network (CSMGAN) is proposed that recasts this task as a process of iterative messages passing over a joint graph that is able to effectively capture high-order interactions between two modalities, thus enabling a further precise localization.
- GuardHealth: Blockchain empowered secure data management and Graph Convolutional Network enabled anomaly detection in smart healthcare142
This paper proposes GuardHealth: an efficient, secure and decentralized Blockchain system for data privacy preserving and sharing, which prevents data sharing without permission and is applicable for smart healthcare system.
- A Survey of Attacks on Large Vision–Language Models: Resources, Advances, and Future Trends130
This article provides a comprehensive review of the various forms of existing LVLM attacks, and systematically review the development of LVLM attack methods, such as adversarial attacks that manipulate model outputs, jailbreak attacks that exploit model vulnerabilities for unauthorized actions, and data poisoning that affects model training.
- Video Graph Transformer for Video Question Answering121
This paper proposes a Video Graph Transformer model for Video Quetion Answering (VideoQA), and shows that VGT can achieve much better performances on VideoQA tasks that challenge dynamic relation reasoning than prior arts in the pretraining-free scenario.
- Robust discrete code modeling for supervised hashing118
This work proposes a novel supervised hashing approach, termed as Robust Discrete Code Modeling (RDCM), which directly learns high-quality discretebinary codes and hash functions by effectively suppressing the influence of unreliable binary codes and potentially noisily-labeled samples.
- Spatio–Temporal Edge Service Placement: A Bandit Learning Approach116
This paper proposes SEEN, a Spatial-temporal Edge sErvice placemeNt algorithm, a novel combinatorial contextual bandit learning problem of an ASP under a limited budget, and proves that the algorithm achieves a sublinear regret bound when it is compared with an Oracle algorithm that knows the exact benefit information.
- 110
- 110
- 110
- 109
- Multi-Modal Cross-Domain Alignment Network for Video Moment Retrieval107
A novel Multi-Modal Cross-Domain Alignment (MMCDA) network to transfer the annotation knowledge from the source domain to the target domain and raises the performance by more than 7% in representative cases, which demonstrates its effectiveness.
- Fine-grained Iterative Attention Network for Temporal Language Localization in Videos106
A Fine-grained Iterative Attention Network (FIAN) that consists of an iterative attention module for bilateral query-video in-formation extraction and a content-oriented localization strategy instead of applying recent anchor-based localization is proposed.
- 105
- The Security and Privacy of Mobile-Edge Computing: An Artificial Intelligence Perspective99
This article comprehensively provides a survey of security and privacy in MEC from the perspective of AI, using European Telecommunications Standards Institute MEC reference architecture as a framework while merging the software-defined network (SDN) and network function virtualization (NFV) to better illustrate a serviceable platform of MEC.
- Memory-Guided Semantic Learning Network for Temporal Sentence Grounding99
A memory-augmented network, called Memory-Guided Semantic Learning Network (MGSL-Net), that learns and memorizes the rarely appeared content in TSG task and substantially improves the accuracy not only on the entire dataset but also on the rare cases.
- Hierarchical Local-Global Transformer for Temporal Sentence Grounding98
A novel Hierarchical Local-Global Transformer (HLGT) is proposed to leverage this hierarchy information and model the interactions between different levels of granularity and different modalities for learning more fine-grained multi-modal representations.
- 98
- 98
- You Can Ground Earlier than See: An Effective and Efficient Pipeline for Temporal Sentence Grounding in Compressed Videos97
A new setting is posed, compressed-domain TSG, which directly utilizes compressed videos rather than fully-decompressed frames as the visual input, and a novel Three-branch Compressed-domain Spatial-temporal Fusion (TCSF) framework, which extracts and aggregates three kinds of low-level visual features for effective and efficient grounding.
- 97
- ANIMC: A Soft Approach for Autoweighted Noisy and Incomplete Multiview Clustering96
Compared with previous methods, ANIMC has three unique advantages: 1) it is a soft algorithm to adjust the approach in different scenarios, thereby improving its generalization ability; 2) it automatically learns a proper weight for each view, thereby reducing the influence of noises; 3) it performs doubly soft regularized regression that aligns the same instances in different views, thereby decreasing the impact of missing instances.
- 95
- PIRATE: A Blockchain-Based Secure Framework of Distributed Machine Learning in 5G Networks94
This article proposes a secure computing framework based on the sharding technique of blockchain, namely PiRATE, which aims to provide the byzantine-resilience for distributed learning in the 5G era.
- High-force soft pneumatic actuators based on novel casting method for robotic applications93
This study helps improve the fabrication process for fiber-reinforced actuators and propose new high-force SPAs for robotic applications, which provides important hardware for the development of high-load soft robots.
- SAANet: Siamese action-units attention network for improving dynamic facial expression recognition93
A metric learning framework with a siamese cascaded structure that learns a fine-grained distinction for different expressions in video-based task and a novel action-units attention mechanism tailored to FER task to extract spatial contexts from the emotion regions is proposed.
- Microneedle-mediated drug delivery for cutaneous diseases87
This review mainly introduces microneedles fabricated with different technologies and transdermal delivery of various therapeutics for cutaneous diseases, such as psoriasis, atopic dermatitis, skin and soft tissue infection, superficial tumors, axillary hyperhidrosis, and plantar warts.
- Optimal Repair Layering for Erasure-Coded Data Centers85
A practical repair layering framework called DoubleR is proposed and implemented atop the Hadoop Distributed File System (HDFS) and it is shown that DoubleR maintains the theoretical guarantees of DRC and improves the repair performance of regenerating codes in both node recovery and degraded read operations.
- 84
- Differentially Private Online Learning for Cloud-Based Video Recommendation With Multimedia Big Data in Social Networks84
A cloud-assisted differentially private video recommendation system based on distributed online learning that can greatly reduce the performance loss and keep a delicate balance between the total reward and privacy preserving level is proposed.
- Jailbreaking GPT-4V via Self-Adversarial Attacks with System Prompts83
This work discovers a system prompt leakage vulnerability in GPT-4V and proposes a novel MLLM jailbreaking attack method termed SASP (Self-Adversarial Attack via System Prompt), demonstrating the important role of system prompts in jailbreaking.
- 81
- LeaNet: Lightweight U-shaped architecture for high-performance skin cancer image segmentation80
The proposed LeaNet is a novel U-shaped network for high-performance yet lightweight skin cancer image segmentation that employs multiple attention blocks in a lightweight symmetric U-shaped design to enhance the model's multi-level feature extraction capability.
- 77
- Adaptive Fog Configuration for the Industrial Internet of Things77
An online distributed algorithm, called adaptive fog configuration (AFC), based on Lyapunov optimization and parallel Gibbs sampling is proposed, which jointly optimizes service hosting and task admission decisions, requiring only currently available system information while guaranteeing close-to-optimal performance compared to an oracle algorithm with full future information.
- Adaptive Proposal Generation Network for Temporal Sentence Localization in Videos76
This paper proposes an Adaptive Proposal Generation Network (APGN) to maintain the segment-level interaction while speeding up the efficiency and significantly outperforms previous state-of-the-art methods on three challenging benchmarks.
- A 3D-CNN and LSTM Based Multi-Task Learning Architecture for Action Recognition76
A novel MTL architecture that first combines 3D convolutional neural networks (3D CNN) plus the long short-term memory (LSTM) networks together with the MTL mechanism, tailored to information sharing of video inputs is proposed.
- Modality Attention for End-to-end Audio-visual Speech Recognition76
A novel multimodal attention based method for audio-visual speech recognition which could automatically learn the fused representation from both modalities based on their importance, realized using state-of-the-art sequence-to-sequence (Seq2seq) architectures.
- 74
- 74
- Using artificial intelligence to rapidly identify microplastics pollution and predict microplastics environmental behaviors73
This review provides a comprehensive overview of machine learning methods for the prediction of MPs for various tasks, and discusses in detail the data source, data preprocessing, algorithm principle, and algorithm limitation of applied machine learning.
- 72
- 70
- 69
- 69
- 69
- 68
- 67
- 66
- 65
- Instant3D: Instant Text-to-3D Generation64
A novel framework for fast text-to-3D generation, dubbed Instant3D, that is able to create a 3D object for an unseen text prompt in less than one second with a single run of a feedforward network.
- 64
- Exploring Optical-Flow-Guided Motion and Detection-Based Appearance for Temporal Sentence Grounding62
A novel Motion- and Appearance-guided 3D Semantic Reasoning Network (MA3SRN) is proposed, which incorporates optical-flow-guided motion-aware, detection-based appearance- Aware, and 3D-aware object-level features to better reason the spatial-temporal object relations for accurately modelling the activity among consecutive frames.
- Unsupervised Temporal Video Grounding with Deep Semantic Clustering62
This paper proposes a novel Deep Semantic Clustering Network (DSCNet) to leverage all semantic information from the whole query set to compose the possible activity in each video for grounding and demonstrates that the DSCNet achieves competitive performance, and even outperforms most weakly-supervised approaches.
- 62
- An untethered soft robotic gripper with high payload-to-weight ratio60
A tightening force model for winding PAMs is presented in this paper, which can be used to evaluate the pressing force of the untethered soft gripper on the surface of an object.
- 58
- Attend to count: Crowd counting with adaptive capacity multi-scale CNNs57
The Adaptive Capacity Multi-scale convolutional neural networks (ACM-CNN), a novel crowd counting approach which can assign different capacities to different portions of the input, is proposed.
- NFV-Enabled Multicasting in Mobile Edge Clouds with Resource Sharing56
A provable approximation algorithm with an approximation ratio for the problem of NFV-enabled multicasting in a mobile edge cloud, where each multicast request requires to process its traffic in a specified sequence of network functions before the traffic from a source to a set of destinations.
- Progressively Guide to Attend: An Iterative Alignment Framework for Temporal Sentence Grounding55
This paper proposes an Iterative Alignment Network (IA-Net) for TSG task, which iteratively interacts inter- and intra-modal features within multiple steps for more accurate grounding and can robustly capture the fine-grained relations between vision and language domains step-by-step for progressively reasoning the temporal boundaries.
- F2Net: Learning to Focus on the Foreground for Unsupervised Video Object Segmentation55
A novel Focus on Foreground Network (F2Net), which delves into the intra-inter frame details for the foreground objects and thus effectively improve the segmentation performance.
- MetaCloak: Preventing Unauthorized Subject-Driven Text-to-Image Diffusion-Based Synthesis via Meta-Learning54
This work solves the bi-level poisoning problem with a meta-learning framework with an additional transformation sampling process to craft transferable and robust perturbation, and employs a pool of surrogate diffusion models to craft transferable and model-agnostic perturbation.
- DA-Net: Learning the Fine-Grained Density Distribution With Deformation Aggregation Network54
The deformation aggregation network (DA-Net) is proposed that can incrementally incorporate adaptive receptive fields to capture the fine-grained density distribution and delivers the state-of-the-art performance on four benchmarks.
- Fewer Steps, Better Performance: Efficient Cross-Modal Clip Trimming for Video Moment Retrieval Using Language53
This paper designs a novel clip search model that learns to identify promising video regions to search conditioned on the language query and introduces a set of low-cost semantic indexing features to capture the context of objects and interactions that suggest where to search the query-relevant moment.
- STPrivacy: Spatio-Temporal Privacy-Preserving Action Recognition53
A novel framework STPrivacy is proposed to perform video-level PPAR and introduces vision Transformers into PPAR by treating a video as a tubelet sequence, and accordingly design two complementary mechanisms, i.e., sparsification and anonymization, to remove privacy from a spatio-temporal perspective.
- 53
- 52
- Pandora's Box: Towards Building Universal Attackers against Real-World Large Vision-Language Models51
This paper makes the first attempt to build a universal attacker against real-world LVLMs, focusing on two critical aspects: restricting access to only the LVLM inputs and outputs and devising a universal adversarial patch, which is task-agnostic and can deceive any LVLM-driven task when applied to various inputs.
- 51
- MHP-VOS: Multiple Hypotheses Propagation for Video Object Segmentation51
A novel approach to defer the decision making for a target object in each frame, until a global view can be established with the entire video being taken into consideration, and a novel mask merging strategy is employed to handle mask conflicts between objects.
- Position-Guided Text Prompt for Vision-Language Pre-Training50
This work proposes a novel Position-guided Text Prompt (PTP) paradigm to enhance the visual grounding ability of cross-modal models trained with VLP, and introduces PTP into several state-of-the-art VLP frameworks.
- 50
- Not All Inputs Are Valid: Towards Open-Set Video Moment Retrieval using Language49
This paper proposes a novel model OpenVMR, which first distinguishes ID and OOD queries based on the normalizing flow technology, and then conducts moment retrieval based on ID queries, and first distinguishes ID and OOD queries based on the normalizing flow technology, and then conducts moment retrieval based on ID queries.
- 48
- 48
- A Bioinspired Soft Swallowing Robot Based on Compliant Guiding Structure48
A novel soft swallowing robot (SSR) capable of swallowing various objects that have different shapes and stiffnesses, protecting objects from squeeze and collision, and withstanding high temperature is proposed, enabled by a compliant guiding structure consisting of a double thin-walled capsule filled with fluid and a linearly movable traction body.
- Periodic Charging for Wireless Sensor Networks With Multiple Portable Chargers48
This paper presents an efficient slot-based periodic charging time scheduling algorithm with both a fine-grained node classification scheme to prevent unnecessary visits of energy-sufficient nodes, and a balanced charging task assignment scheme to avoid charging starvation.
- Differentially Private Federated Learning on Non-iid Data: Convergence Analysis and Adaptive Optimization47
This work is the first to adopt the truncated concentrated differential privacy technique under the FL scenario to more tightly track end-to-end privacy loss, while requiring less noise injection for the same level of DP.
- Video-based Facial Expression Recognition using Graph Convolutional Networks47
A Graph Convolutional Network layer is introduced into a common CNN-RNN based model for video-based FER task, and it is the first time to use GCN in FER, and the experimental results demonstrate that the method has superior performance to existing methods.
- 47
- An Online Attention-Based Model for Speech Recognition47
An adaptive monotonic chunk-wise attention (AMoChA) mechanism is proposed to replace GSA for the calculation of attention weight distribution and two methods to alleviate the huge performance degradation when combining LC and AMo ChA are proposed.
- 45
- Privacy-preserving and energy efficient task offloading for collaborative mobile computing in IoT: An ADMM approach45
This paper proposes an energy efficient task offloading problem subject to the overall task delay based on Alternating Direction Method of Multipliers (ADMM) in a three-tier MEC network, equipped with both edge nodes and the cloud and addresses the privacy disclosure concerns in the data transmission among IoT devices.
- 45
- 43
- Rethinking Weakly-Supervised Video Temporal Grounding From a Game Perspective42
This is the first attempt to tackle the challenging task of weakly-supervised video temporal grounding from a novel game perspective, which effectively learns the uncertain relationship between each vision-language pair with diverse granularity and flexible combination for multi-level cross-modal interaction.
- 42
- ScaleLong: Towards More Stable Training of Diffusion Model via Scaling Network Long Skip Connection41
An effective coefficient scaling framework ScaleLong is proposed that scales the coefficients of LSC in UNet and better improves the training stability of UNet.
- 41
- 41
- 41
- Coarse to Fine: Domain Adaptive Crowd Counting via Adversarial Scoring Network40
A novel adversarial scoring network (ASNet) to gradually bridge the gap across domains from coarse to fine granularity, and designs a dual-discriminator strategy to adapt source domain to be close to the targets from the perspectives of both global and local feature space via adversarial learning.
- Reasoning Step-by-Step: Temporal Sentence Localization in Videos via Deep Rectification-Modulation Network40
A novel deep rectification-modulation network (RMN) is proposed, transforming this task into a multi-step reasoning process by repeating rectification and modulation, thus enabling a further precise localization of temporal sentence localization in videos.
- 40
- Rethinking the Video Sampling and Reasoning Strategies for Temporal Sentence Grounding39
A novel Siamese Sampling and Reasoning Network (SSRN) for TSG is proposed, which introduces a siamese sampling mechanism to generate additional contextual frames to enrich and refine the new boundaries.
- 39
- Adversarial Meta Sampling for Multilingual Low-Resource Speech Recognition39
This work develops a novel adversarial meta sampling (AMS) approach to improve MML-ASR, and demonstrates the applicability of AMS to other low-resource speech tasks and transfer learning ASR approaches.
- Joint Offloading and Charge Cost Minimization in Mobile Edge Computing39
An iterative algorithm is proposed to obtain a suboptimal solution in polynomial time for downlink resource allocation, offloading decision and computation resource allocation and results show that the proposed algorithm performs better than benchmark algorithms.
- Annotations Are Not All You Need: A Cross-modal Knowledge Transfer Network for Unsupervised Temporal Sentence Grounding38
By modulating and transferring both appearance and action knowledge into the authors' challenging unsupervised task, the model can directly utilize this general knowledge to correlate videos and queries, and accurately retrieve the relevant segment without training.
- 38
- 37
- 37
- 37
- 3DHacker: Spectrum-based Decision Boundary Generation for Hard-label 3D Point Cloud Attack35
A novel 3D attack method, termed 3D Hard-label attacker (3DHacker), based on the developed decision boundary algorithm to generate adversarial samples solely with the knowledge of class labels, which competitively outperforms existing 3D attacks regarding the attack performance as well as adversary quality.
- 35
- Exploring RNN-Transducer for Chinese speech recognition35
This paper explores RNN-T for a Chinese large vocabulary continuous speech recognition (LVCSR) task and proposes a new strategy of learning rate decay and finds that adding convolutional layers at the beginning of the network and using ordered data can discard the pre-training process of the encoder without loss of performance.
- Private and Truthful Aggregative Game for Large-Scale Spectrum Sharing35
A novel game called aggregative game is introduced and used to model spectrum sharing in large-scale, heterogeneous, and dynamic networks and it is shown that the learning algorithm achieves an NE over time and provides no regret guarantee for each user.
- 34
- 34
- Differentially Private Triangle Counting in Large Graphs34
This paper publishes the triangle counts satisfying the node-differential privacy with two kinds of histograms: the triangle count distribution and the cumulative distribution, and proposes a novel graph projection method that can be used to obtain an upper bound for sensitivity in different distributions.
- 34
- Zeroth-Order Fine-Tuning of LLMs in Random Subspaces33
This paper introduces a low-rank perturbation tailored for LLMs that significantly reduces memory consumption while improving performance, and proves that the gradient estimation closely approximates the backpropagation gradient, exhibits lower variance than traditional ZO methods, and ensures convergence when combined with SGD.
- 33
- 33
- MLCA-AVSR: Multi-Layer Cross Attention Fusion Based Audio-Visual Speech Recognition32
This study proposes a multi-layer cross-attention fusion based AVSR (MLCA-AVSR) approach that promotes representation learning of each modality by fusing them at different levels of audio/visual encoders.
- 32
- 31
- 30
- Efficient secure and verifiable location-based skyline queries over encrypted data30
This paper proposes an authenticated data structure, named secure and verifiable scope R-tree (SVSR-tree), and develops several secure protocols based on SVSR- tree to accelerate the query efficiency and reduce the size of verification objects.
- 30
- From Spatial to Actions: Grounding Vision-Language-Action Model in Spatial Foundation Priors29
This work introduces FALCON (From Spatial to Action), a novel paradigm that injects rich 3D spatial tokens into the action head and includes an Embodied Spatial Model that can optionally fuse depth, or pose for higher fidelity when available, without retraining or architectural changes.
- 29
- 29
- 29
- 29
- 29
- 28
- 28
- Efficient, Direct, and Restricted Black-Box Graph Evasion Attacks to Any-Layer Graph Neural Networks via Influence Function27
This work introduces two influence functions, i.e., feature-label influence and label influence, that are defined on GNNs and label propagation (LP), respectively, and proposes an efficient algorithm to calculate label influence.
- Sparse Enhanced Network: An Adversarial Generation Method for Robust Augmentation in Sequential Recommendation27
A Sparse Enhanced Network (SparseEnNet) is proposed, which is a robust adversarial generation method, allowing the model to differentiate between data augmentation categories and achieve better prediction performance for the next item in the sequence.
- 27
- 26
- 26
- 26
- 26
- 25
- Design and modeling of a high-load soft robotic gripper inspired by biological winding25
This work proposes a high-load soft gripper driven by pneumatic artificial muscles (PAMs) that combines the advantages of a high force soft actuator and a closed gripping structure and contributes to the load improvement of soft robotic grippers, and the mathematical modeling of engineering systems with winding structures.
- 25
- 24
- 24
- What Makes Good Collaborative Views? Contrastive Mutual Information Maximization for Multi-Agent Perception22
This paper proposes a novel framework named CMiMC (Contrastive Mutual Information Maximization for Collaborative Perception) based on multi-view contrastive learning to realize estimation and maximization of MVMI, which assists the training of a collaborative encoder for voxel-level feature fusion.
- Bandits for Structure Perturbation-based Black-box Attacks to Graph Neural Networks with Theoretical Guarantees22
This paper proposes an online attack based on bandit optimization which is proven to be sublinear to the query number T, i.e., O(✓NT3/4) where N is the number of nodes in the graph.
- 22
- 22
- 22
- 22
- 21
- 21
- 21
- 21
- 21
- BadToken: Token-level Backdoor Attacks to Multi-modal Large Language Models20
This work proposes BadToken, the first token-level backdoor attack to MLLMs, and introduces two novel backdoor behaviors: Token-substitution and Token-addition, which enable flexible and stealthy attacks by making token-level modifications to the original output for backdoored inputs.
- 20
- 20
- DPlis: Boosting Utility of Differentially Private Deep Learning via Randomized Smoothing20
The core idea of DPlis is to construct a smooth loss function that favors noise-resilient models lying in large flat regions of the loss landscape that can effectively boost model quality and training stability under a given privacy budget.
- De-Health: All Your Online Health Information Are Belong to Us20
A novel online health data De-Anonymization (DA) framework, named De-Health, is presented and the DA efficacy is validated, and a linkage attack framework is presented which can link online health/medical information to real world people.
- 20
- 20
- 20
- Manifold Constraints for Imperceptible Adversarial Attacks on Point Clouds19
A bijective manifold mapping between point clouds and a simple parameter shape is constructed using an invertible auto-encoder, so that manifold-aware distortion during attacks can be captured within the parameter space.
- Phenotyping of poplar seedling leaves based on a 3D visualization method19
The results proved that this approach could be used to accurately measure the leaf characteristics of poplar seedlings via visualization, and NURBS was used to fit the leaf surface based on non-uniform rational B-splines.
- AutoJailbreak: Exploring Jailbreak Attacks and Defenses through a Dependency Lens18
A systematic analysis of the dependency relationships in jailbreak attack and defense techniques, generalizing them to all possible attack surfaces and introducing a novel evaluation method that distinguishes hallucinations, which are often overlooked, from jailbreak attack and defense responses.
- 18
- 18
- 18
- 18
- 18
- On heterogeneous duty cycles for neighbor discovery in wireless sensor networks18
Two neighbor discovery protocols are proposed, called Hedis and Todis, that control the duty cycle granularity of quorum and co-primality based protocols respectively, by enabling the finest-grained control of heterogeneous duty cycles.
- Tree-Based Contextual Learning for Online Job or Candidate Recommendation With Big Data Support in Professional Social Networks18
An online mining and predicting system is proposed for personalized job or candidate recommendation with big-data support that considers the users’ explicit information as context as well as online implicit information of the previous users to achieve personalized recommendation.
- 18
- 18
- 17
- 17
- 17
- Differentially Private Deep Learning with Iterative Gradient Descent Optimization17
A novel perturbed iterative gradient descent optimization (PIGDO) algorithm is proposed and it is proved that this algorithm satisfies the differential privacy and a modified moments accountant (MMA) method is proposed to conduct the privacy analysis and obtain a tighter bound of privacy loss compared with the original moments accountant method.
- 17
- 17
- 17
- 17
- Filling the Information Gap between Video and Query for Language-Driven Moment Retrieval16
A Teacher-Student Moment Retrieval framework to fill the cross-modal information gap during the moment retrieval guided by a single query and introduces an adaptive knowledge distillation mechanism to train a student model with a single queries input by selectively absorbing the knowledge from the teacher model.
- RBNN: Memory-Efficient Reconfigurable Deep Binary Neural Network With IP Protection for Internet of Things16
This article proposes a reconfigurable BNN (RBNN) to further amplify the memory efficiency for resource-constrained Internet of Things (IoT) devices while naturally protecting the model IP.
- 16
- 16
- 16
- 16
- 16
- Way to Specialist: Closing Loop Between Specialized LLM and Evolving Domain Knowledge Graph15
The Way-to-Specialist (WTS) framework is proposed, which synergizes retrieval-augmented generation with knowledge graphs (KGs) to enhance the specialized capability of LLMs in the absence of specialized training.
- 15
- Video Content Placement At the Network Edge: Centralized and Distributed Algorithms15
This paper uses the Lyapunov optimization technique to formulate the video content placement problem as a series of one-time-slot optimization problems and applies an Alternating Direction Method of Multipliers (ADMM)-based method to solve each of them.
- Joint time delay and energy optimization with intelligent overclocking in edge computing15
An intelligent overclocking mechanism for the MEC server that operates for an intelligently calculated period to allow it to leverage more computing power without introducing additional hardware resources for a certain period of time is suggested.
- 15
- Generating Robust Audio Adversarial Examples with Temporal Dependency15
A new Iterative Proportional Clipping (IPC) algorithm is proposed that preserves temporal dependency in audios for generating more robust adversarial examples and can significantly reduce human-perceptible noises and resist the defenses based on the temporal structure.
- 15
- 15
- Dynamic online convex optimization with long-term constraints via virtual queue14
This work proposes a novel iterative optimization algorithm based on the virtual queues to achieve sublinear regret and constraint violations in the online convex optimization (OCO) with long-term constraints.
- 14
- Reading Digital Numbers of Water Meter with Deep Learning Based Object Detector14
In order to attack the half-character problem of water meter, this work proposed a heuristic rule to guarantee that there is only one bounding box in the vertical direction within a grid, which outperforms the traditional water meter recognition system in both time and precision.
- 14
- 13
- 13
- Transform-Equivariant Consistency Learning for Temporal Sentence Grounding13
A novel Equivariant Consistency Regulation Learning (ECRL) framework is introduced to learn more discriminative query-related frame-wise representations for each video, in a self-supervised manner, to address the temporal sentence grounding.
- Backdoor Attacks to Pre-trained Unified Foundation Models13
This paper makes proof-of-concept level research on the backdoor attack for pre-trained unified foundation models and reveals the vulnerability of these models and suggests future research directions for enhancing the attack approach.
- 13
- 13
- A Case Study of HealthCare Platform using Big Data Analytics and Machine Learning13
The experiment result shows the high accuracy over 95% of the disease detection using machine learning methods, with the cost 90% lower than the local hospital in Bangladesh, which provides the strong support to implement of the platform in the remote area of the country.
- 13
- 13
- 12
- 12
- Development of a Multi-Cable-Driven Continuum Robot Controlled by Parallel Platforms12
This work proposes a method for increasing the number of cables without increasing actuators in a continuum robot through parallel platforms, and develops a prototype of six-cable-driven continuum robot controlled by 3RPS parallel platforms.
- 12
- 11
- Divide, Conquer, and Coalesce: Meta Parallel Graph Neural Network for IoT Intrusion Detection at Scale11
An offline reinforcement learning algorithm, called Graph-Embedded Adversarially Trained Actor-Critic (G-ATAC), is proposed to learn a coalition formation policy that jointly optimizes intrusion detection accuracy, communication overheads, and computational complexities of coalitional NIDSs.
- 11
- 11
- 11
- Exploring Deep Reinforcement Learning for Task Dispatching in Autonomous On-Demand Services11
A novel demand dispatching model using deep reinforcement learning that outperforms the existing models in terms of Cancellation Rate and Average Response Time and is applicable to autonomous on-demand services.
- 11
- DP-Hybrid: A Two-Layer Consensus Protocol for High Scalability in Permissioned Blockchain11
DP-Hybrid is proposed, a novel two-layer consensus protocol that reduces the communication costs from PBFT’s O ( N 2 ) to O ( N 2 /K 2 ).
- 11
- 11
- Towards Scalable and Consistent 3D Editing10
3DEditFormer, a 3D-structure-preserving conditional transformer that disentangles editable regions from preserved structure, enabling precise and consistent edits without requiring auxiliary 3D masks is proposed.
- 10
- 10
- 10
- 10
- U2-KWS: Unified Two-Pass Open-Vocabulary Keyword Spotting with Keyword Bias10
A novel unified two-pass open-vocabulary KWS (U2-KWS) framework inspired by the two-pass ASR model U2, which can achieve a significant relative wake-up rate improvement compared to the traditional customized KWS systems when the false alarm rate is fixed to 0.5 times per hour.
- 10
- 10
- 10
- 10
- 10
- 10
- Scrub It Out! Erasing Sensitive Memorization in Code Language Models via Machine Unlearning9
This paper conducts a pioneering investigation into erasing sensitive memorization in CLMs through machine unlearning—a post-hoc modification method that removes specific information from trained models without requiring full retraining.
- 9
- Effective and Efficient Adversarial Detection for Vision-Language Models via A Single Vector9
A novel and effective iN-time Embedding-based AdveRSarial Image DEtection (NEARSIDE) method, which exploits a single vector that distilled from the hidden states of VLMs, which is called the attacking direction, to achieve the detection of adversarial images against benign ones in the input.
- Learning-driven service caching in MEC networks with bursty data traffic and uncertain delays9
This work proposes an online learning algorithm for the problem with uncertain processing delays, by utilizing the multi-armed bandits technique, and analyze the regret bound of the proposed algorithm, and devise an efficient heuristic for the problem with the uncertainties of both bursty data traffic and uncertain delays.
- 9
- An Improved Multilabel k-Nearest Neighbor Algorithm Based on Value and Weight9
An improved ML-kNN algorithm based on value and weight is proposed that could be applied to a multilabel classification of various fields involving label imbalance, such as drug molecule identification, building identification, and text categorization.
- 9
- Time-aspect-sentiment Recommendation Models Based on Novel Similarity Measure Methods9
This article collects the description information and the reviews of the items from public websites, then adopts sentiment analysis techniques to model the similarities on user level and item level, and proposes Item-User Level Similarity Matrix Factorization (IULMF), which combines these two methods to study their contributions on the final performance.
- k-Reciprocal Harmonious Attention Network for Video-Based Person Re-Identification9
A reciprocal harmonious attention network to jointly learn discriminative spatiotemporal features and the similarity metrics and the performance of the proposed approach outperforms the state-of-the-art methods.
- 8
- ICMC-ASR: The ICASSP 2024 In-Car Multi-Channel Automatic Speech Recognition Challenge8
The ICASSP 2024 In-Car Multi-Channel Automatic Speech Recognition Challenge is launched and first-place team USTC-iflytek achieves a CER of 13.16% in the ASR track and a cpCER of 21.48% in the ASDR track, showing an absolute improvement of 13.08% and 51.4% compared to the challenge baseline, respectively.
- 8
- 8
- Reversible Data Hiding Based on Structural Similarity Block Selection8
This paper proposes a new RDH method to concentrate on both PSNR and SSIM performance, which involves two key techniques: an SSIM-based block selection technique, and a PEH-based optimal expansion bins selection technique.
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- Hard-Label Black-Box Attacks on 3D Point Clouds7
This work introduces a novel 3D attack method based on a new spectrum-aware decision boundary algorithm to generate high-quality adversarial samples, which competitively outperforms existing white/black-box attackers in terms of attack performance and adversary quality.
- Jointly Visual- and Semantic-Aware Graph Memory Networks for Temporal Sentence Localization in Videos7
A novel Hierarchical Visual- and Semantic-Aware Reasoning Network (HVSARN) is proposed, which enables both visual- and semantic-aware query reasoning from object-level to frame-level and a new graph memory mechanism to perform visual-semantic query reasoning.
- Mountainous village relocation planning with 3D GIS virtual imaging space model and neural network7
The results demonstrate that compared with the traditional survey method, the proposed method of digital survey can better address the issue of remote mountain areas, including provide more accurate, complete and sustainable information and make the surveying and planning more convenient, efficient and smart.
- 7
- 7
- 7
- On Cooperative Channel Rendezvous in Cognitive Radio Networks7
This paper develops a novel cooperative rendezvous mechanism by introducing the cooperative cognitive nodes serving as “bridges” between communicating pairs to facilitate their rendezvous process, and establishes a mathematics framework to study the rendezvous delay and robustness and derive the performance limit in asymptotic scenarios.
- 7
- Optimal coverage configuration based on artificial fish swarm algorithm in WSNs7
Simulation results show that the algorithm can get an optimal selection of the set of nodes and improve the real-time capability of nodes scheduling, and proposed a coverage optimization strategy based on fish swarm algorithm.
- 7
- 6
- Genetic Auto-prompt Learning for Pre-trained Code Intelligence Language Models6
GenAP is the first work to automatically design prompts for code intelligence PLMs, which utilizes an elaborate genetic algorithm to automatically design prompts and outperforms all other methods across all three tasks.
- 6
- 6
- Unlearnable Graph: Protecting Graphs from Unauthorized Exploitation6
A novel method for generating unlearnable graph examples by injecting delusive but imperceptible noise into graphs using the authors' Error-Minimizing Structural Poisoning (EMinS) module, which is able to make the graphs unexploitable.
- 6
- 6
- 6
- 6
- 6
- 6
- Cost-Efficient and Skew-Aware Data Scheduling for Incremental Learning in 5G Networks6
This work proposes Cocktail, an incremental learning framework within a reference 5G network architecture that exploits the stochastic gradient descent to devise an online asymptotically optimal algorithm, including two optimal policies based on novel graph constructions for skew-aware data collection and data training.
- 6
- 6
- 6
- Identity-Aware Attribute Recognition via Real-Time Distributed Inference in Mobile Edge Clouds6
Evaluation results show that the performance of the proposed algorithm with distributed inference framework is promising, by reaching the accuracies of attribute recognition and person identification up to 92.9% and 96% respectively, and significantly reducing the inference delay by at least 40.6% compared with existing methods.
- 6
- Understanding Deep Representations through Random Weights.6
This work reveals that the purely random CNN-DeCNN architecture substantially contributes to the geometric and photometric invariance due to the intrinsic symmetry and invertible structure, but it discards the colormetric information due toThe random projection.
- 6
- 5
- 5
- 5
- 5
- GraphCloak: Safeguarding Task-specific Knowledge within Graph-structured Data from Unauthorized Exploitation5
GraphCloak is introduced to safeguard against the unauthorized usage of graph data by proposing two error-minimizing-based poisoning methods that target perturbations on the structural and feature space, along with a subgraph injection poisoning method.
- 5
- 5
- 5
- 4
- 4
- Towards Natural Image Matting in the Wild via Real-Scenario Prior4
This work proposes SEMat, a new matting dataset based on the COCO dataset, namely COCO-Matting, which revamps the network architecture and training objectives and proves its efficacy in interactive natural image matting.
- An Adaptive Harmony Search Part-of-Speech tagger for Square Hmong Corpus4
A weight evaluation function to reduce the influence of unknown words is designed utilizing the roulette and local evaluation strategies for handling the square Hmong part-of-speech tagging problem.
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- Cross-lingual sentiment analysis for low-resource languages via semantic alignment and transfer learning3
This work advances multilingual NLP by providing a scalable solution for low-resource language processing by proposing a cross-lingual transfer augmentation method (CTAM) that integrates semantic alignment and adversarial training.
- 3
- 3
- 3
- Exploring the Robustness of Decentralized Training for Large Language Models3
The importance of addressing security concerns in the context of decentralized training for large language models and the fundamental difference between decentralized foundation model training and vanilla federated learning is highlighted.
- Dynamical analysis for a diffusive SVEIR epidemic model with nonlinear incidences3
The well-posedness of solutions for the model is obtained and the disease-free equilibrium is globally asymptysym, which shows that the persistence theory of dynamical systems, the authors show, is globally persistent.
- 3
- ATPGNN: Reconstruction of Neighborhood in Graph Neural Networks With Attention-Based Topological Patterns3
This article proposes a semi-supervised learning platform ATPGNN with three variable components that can fully adapt to different tasks by changing its components and support inductive learning.
- 3
- 3
- 3
- 3
- SNEAK: Synonymous Sentences-Aware Adversarial Attack on Natural Language Video Localization2
A new adversarial attack paradigm called synonymous sentences-aware adversarial attack on NLVL (SNEAK), which captures the cross-modality interplay between the vision and language sides is proposed, which captures the cross-modality interplay between the vision and language sides.
- 2
- Towards Robust Overlapping Speech Detection: A Speaker-Aware Progressive Approach Using WavLM2
This work proposes a speaker-aware progressive OSD model that leverages a progressive training strategy to enhance the correlation between subtasks such as voice activity detection (VAD) and overlap detection.
- Deep Learning-Based FSS Spectral Characterization and Cross-Band Migration2
This paper systematically analyzes the impact of material and geometric parameters of FSSs on their spectral characteristics, thereby establishing a theoretical foundation for the cross-band transfer learning capability of neural networks and demonstrates the feasibility of transfer learning.
- Graph Agent Network: Empowering Nodes with Inference Capabilities for Adversarial Resilience2
Experimental results show that GAgN effectively implements all its intended capabilities and, compared to state-of-the-art defenses, achieves optimal classification accuracy on the perturbed datasets, and it is proved that single-hidden-layer multilayer perceptrons (MLPs) are theoretically sufficient to achieve these functionalities.
- 2
- Optimal control analysis in a reaction-diffusion SIRC model with cross-immune class2
This paper proves the existence and uniqueness of the global positive strong solution to the control system by employing the theories of semigroups of operators and establishes the first-order necessary optimality conditions that the optimal control must satisfy, employing the methods of convex perturbation.
- 2
- 2
- 2
- 2
- FedNow: An Efficiency-Aware Incentive Mechanism Enables Privacy Protection and Efficient Resource Utilization in Federated Edge Learning2
An efficiency-aware incentive mechanism, FedNow, is proposed, which enables edge servers to personally determine their local training rounds while motivating their participation without giving access to their true training strategies and private attributes.
- Extracting actors and use cases from requirements text with BiLSTM-CRF2
This study proposes a model for extracting actors and using cases automatically, which combines bi-directional long short-term memory (BiLSTM) and conditional random fields and CRF, and results show that the accuracy of extraction is significantly improved compared with the baseline method.
- Least squares support vector machine method for load identification of nonlinear system2
Simulation results reveal that least squares support vector machine is able to identify reliable inverse model of nonlinear system and then reconstruct accurate real time excitation force.
- 2
- 2
- 2
- Differentially Private Online Learning for Video Recommendation with Social Big Data over Media Cloud.2
A cloud-assisted differentially private video recommendation system based on distributed online learning that can greatly reduce the performance (recommendation accuracy) loss and keep a delicate balance between accuracy and privacy preserving level is proposed.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- AHP Model Construction for Monitoring and Feedback of Australian Bush Fires Data Based on UAV Patrol1
- 1
- End-to-end contextual asr based on posterior distribution adaptation for hybrid ctc/attention system1
- Research on Safe Motion Tracking Algorithm of Manipulator Based on Minimum Pseudo-Distance Criterion1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- Imbalance Rectification in Deep Logistic Regression for Multi-Label Image Classification Using Random Noise Samples1
This work finds that feeding randomly generated noise samples into an LR classifier is an effective way to detect class imbalances, and defines an informative imbalance metric named inference tendency based on noise sample analysis.
- Robust Caching Control in Crowdsourced Content-Centric Mobile Networking1
This paper designs a novel crowdsourced CCMN framework, uses Kullback-Leibler divergence to characterize user mobility distribution uncertainty and formulate the caching control problem into a chance-constrained robust optimization, and proposes a robust caching control scheme.
- 1
- 1
- 1
- 1
- –
- How Human Experts Educate Specialized LLMs: Filling Knowledge Gaps in KG-Augmented Generation through Hallucination Detection–
Epistemic Cognition-enhanced Specialized LLMs (EC-spLLM), a novel evolutionary framework that instills epistemic cognition into the LLM to systematically exploit both its internal knowledge and expert knowledge to reduce dependence on experts and bridging internal and external knowledge sources to enhance performance.
- Augmenting Cross-View Geo-Localization with Spatial Semantics from Vision Foundation Models–
This work reformulates CVGL from a spatial perspective and proposes an auxiliary spatial semantic alignment (SSA) task that learns spatial structure via vision foundation models (VFM) and BEV transformation to enhance the primary CVGL task.
- –
- Every Locale Has Its Rhythm: Endow Time-Series Transformers with Spatial Insights for Long-term Spatio-Temporal Forecasting–
Graph Time-Series Transformer (GïT), a novel approach aimed at enhancing long-term spatio-temporal forecasting, is introduced, which outperforms existing SOTA in terms of forecasting accuracy.
- –
- –
- –
- –
- –
- –
- –
- –
- VCE: A Zero-Cost Hallucination Mitigation Method of LVLMs via Visual Contrastive Editing–
Visual Contrastive Editing (VCE), a novel post-hoc method that identifies and suppresses hallucinatory tendencies by analyzing the model's response to contrastive visual perturbations, is proposed.
- –
- –
- –
- –
- –
- ReAlign: Bilingual Text-to-Motion Generation via Step-Aware Reward-Guided Alignment–
This work proposes BiHumanML3D, a novel bilingual human motion dataset, which establishes a crucial benchmark for bilingual text-to-motion generation models and proposes a Bilingual Motion Diffusion model (BiMD), which leverages cross-lingual aligned representations to capture semantics, thereby achieving a unified bilingual model.
- –
- –
- –
- –
- –
- –
- [Genetic analysis of a Chinese pedigree with 18q21.2-q22.3 duplication and deletion in two offspring respectively resulting from a maternal intrachromosomal insertion].–
The intrachromosomal insertion carried by the pregnant woman had probably given rise to the 18q21.2-q22.3 duplication and deletion in the two offspring, and above finding has provided a basis for genetic counseling for this pedigree.
- –
- F$^2$AT: Feature-Focusing Adversarial Training via Disentanglement of Natural and Perturbed Patterns–
- A Flexible Escape Skin Bioinspired by the Defensive Behavior of Shedding Scales–
This study proposes a flexible escape skin by attaching artificial scales to a flexible film and enables active control of escape force and skin hardness by changing temperature, increasing the adaptability of the escape skin to the surrounding.
- –
- –
- –
- –
- –
- –
- Research on network traffic acquisition based on in-band telemetry technology–
This paper proposes a method that is fine-grained network traffic parameter measurement based on in-band network telemetry technology and simulates the network environment and attack environment in the simulation environment to obtain the data set required for the experiment.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Optimized Selection of Streaming Servers with GeoDNS for CDN-Delivered Live Streaming–
A livestreaming specific GeoDNS design is proposed for selecting optimal Origin Servers to serve Edge Servers, which can save transit cost for CDNs and achieve better end-viewer's experience.
- –
- Pattern Recognition of Impact in Passenger Car Bumper Using Smart Materials–
The application possibility for the method of discriminable pattern recognition whether impact-object is human-like or not by means of neural networks using smart PZT materials is confirmed and enough pattern recognition indices can be developed and then used to recognize the impact information by two different neural networks.
- –
- –
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 2026-10-12. 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.