Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks461 from public data
- Deep learning for sensor-based activity recognition: A survey2,000
The recent advance of deep learning based sensor-based activity recognition is surveyed from three aspects: sensor modality, deep model, and application and detailed insights on existing work are presented and grand challenges for future research are proposed.
- Generalizing to Unseen Domains: A Survey on Domain Generalization1,906
This paper provides a formal definition of domain generalization and discusses several related fields, and categorizes recent algorithms into three classes: data manipulation, representation learning, and learning strategy, and presents several popular algorithms in detail for each category.
- FedHealth: A Federated Transfer Learning Framework for Wearable Healthcare999
FedHealth is proposed, the first federated transfer learning framework for wearable healthcare that performs data aggregation through federated learning, and then builds relatively personalized models by transfer learning.
- Weighted extreme learning machine for imbalance learning743
A weighted ELM which is able to deal with data with imbalanced class distribution while maintain the good performance on well balanced data as unweighted ELM and generalized to cost sensitive learning.
- 710
- Visual Domain Adaptation with Manifold Embedded Distribution Alignment657
This paper proposes a Manifold Embedded Distribution Alignment (MEDA) approach, which learns a domain-invariant classifier in Grassmann manifold with structural risk minimization, while performing dynamic distribution alignment to quantitatively account for the relative importance of marginal and conditional distributions.
- Balanced Distribution Adaptation for Transfer Learning589
This paper proposes a novel transfer learning approach, named as Balanced Distribution Adaptation (BDA), which can adaptively leverage the importance of the marginal and conditional distribution discrepancies, and several existing methods can be treated as special cases of BDA.
- Transfer Learning with Dynamic Adversarial Adaptation Network401
This paper proposes a novel Dynamic Adversarial Adaptation Network (DAAN) to dynamically learn domain-invariant representations while quantitatively evaluate the relative importance of global and local domain distributions.
- Unsupervised Deep Anomaly Detection for Multi-Sensor Time-Series Signals368
A novel deep learning-based anomaly detection algorithm called CAE-M, which builds a Deep Convolutional Autoencoder to characterize spatial dependence of multi-sensor data with a Maximum Mean Discrepancy (MMD) and a Memory Network consisting of linear and non-linear predictions to capture temporal dependence from time-series data.
- 299
- A Pervasive Approach to EEG‐Based Depression Detection288
This study proves the feasibility of a pervasive three-electrode EEG acquisition system for depression diagnosis and suggested that the absolute power of the theta wave might be a valid characteristic for discriminating depression.
- 264
- Stratified Transfer Learning for Cross-domain Activity Recognition253
The proposed framework, referred to as Stratified Transfer Learning (STL), can dramatically improve the classification accuracy for cross-domain activity recognition and extensively investigate the performance of STL across different degrees of similarities and activity levels between domains.
- A Chinese sign language recognition system based on SOFM/SRN/HMM173
This paper uses the self-organizing feature maps (SOFM) as different signers' feature extractor for continuous hidden Markov models (HMM) so as to transform input signs into significant and low-dimensional representations that can be well modeled by the emission probabilities of HMM.
- Mitigating Herding in Hierarchical Crowdsourcing Networks171
A Lyapunov optimization based decision support approach - the Reputation-aware Task Sub-delegation approach with dynamic worker effort Pricing (RTS-P) - with objective functions aiming to achieve superlinear time-averaged collective productivity in an HCN.
- Easy Transfer Learning By Exploiting Intra-Domain Structures168
This paper proposes a practically Easy Transfer Learning (EasyTL) approach which requires no model selection and hyperparameter tuning, while achieving competitive performance and satisfies the Occam's Razor principle.
- Cross-people mobile-phone based activity recognition161
This work proposes an algorithm known as TransEMDT (Transfer learning EMbedded Decision Tree) that integrates a decision tree and the k-means clustering algorithm for personalized activity-recognition model adaptation and shows that this algorithm outperforms several traditional baseline algorithms.
- Near-infrared fluorescence-based multiplex lateral flow immunoassay for the simultaneous detection of four antibiotic residue families in milk158
A novel near-infrared fluorescence based multiplex lateral flow immunoassay by conjugating a near- Infrared label to broad-specificity monoclonal antibody/receptor as detection complexes is developed for rapidly screening the four families of antibiotic residues in milk.
- GeneCompass: deciphering universal gene regulatory mechanisms with a knowledge-informed cross-species foundation model154
GeneCompass is developed, the first knowledge-informed, cross-species foundation model pre-trained on an extensive dataset of over 120 million single-cell transcriptomes from human and mouse, which marks a milestone in advancing knowledge of universal gene regulatory mechanisms and accelerating the discovery of key cell fate regulators and candidate targets for drug development.
- Domain-invariant Feature Exploration for Domain Generalization149
It is argued that domain-invariant features should be originating from both internal and mutual sides, and DIFEX for Domain-Invariant Feature EXploration is proposed, which achieves state-of-the-art performance.
- Survey on Knowledge Distillation for Large Language Models: Methods, Evaluation, and Application148
A thorough survey from three aspects: method, evaluation, and application, exploring knowledge distillation techniques tailored specifically for LLMs is presented, dividing the methods into white-box KD and black-box KD to better illustrate their differences.
- Deep Transfer Learning for Cross-domain Activity Recognition148
An effective Unsupervised Source Selection algorithm for Activity Recognition (USSAR) that is able to select the most similar K source domains from a list of available domains is proposed and an effective Transfer Neural Network to perform knowledge transfer for Activity recognition (TNNAR).
- Development and optimization of a multiplex lateral flow immunoassay for the simultaneous determination of three mycotoxins in corn, rice and peanut136
A multiplex lateral flow immunoassay developed for the simultaneous on-site determination of three mycotoxins (aflatoxin B1, zearalenone and ochratoxin A) in corn, rice and peanut demonstrated that it can be used for routine monitoring of mycotoxin contamination.
- 133
- Adaptive Memory Networks With Self-Supervised Learning for Unsupervised Anomaly Detection130
A novel approach called Adaptive Memory Network with Self-supervised Learning (AMSL) is proposed to address challenges and enhance the generalization ability in unsupervised anomaly detection and is significantly improves the performance compared to other state-of-the-art methods.
- Rapid Enzyme-Linked Immunosorbent Assay and Colloidal Gold Immunoassay for Kanamycin and Tobramycin in Swine Tissues128
Two ELISA and strip test assays can complement each other for rapid detection of residual Kan and Tob in swine tissues and the effect of pH and antibody amount for gold-antibody conjugation on the strip test sensitivity was determined followed by a theoretical explanation of the effects.
- Dealing with Label Quality Disparity in Federated Learning121
The results show that FOCUS effectively identifies clients with noisy labels and reduces their impact on the model performance, thereby significantly outperforming existing FL approaches.
- A novel random forests based class incremental learning method for activity recognition120
An effective class incremental learning method, named Class Incremental Random Forests (CIRF), to enable existing activity recognition models to identify new activities and is able to recognize new class data continuously with a better performance.
- MetaFed: Federated Learning Among Federations With Cyclic Knowledge Distillation for Personalized Healthcare117
A novel framework called MetaFed is proposed to facilitate trustworthy FL between different federations, which obtains a personalized model for each federation without a central server via the proposed cyclic knowledge distillation.
- A Calibration-Free Localization Solution for Handling Signal Strength Variance106
The results show that the proposed solution solves the signal strength variance problem without extra calibration on test device and performs equally to that of existing calibration-based method.
- 104
- Personalized Federated Learning With Adaptive Batchnorm for Healthcare102
FedAP is proposed to tackle domain shifts and then obtain personalized models for local clients by learning the similarity between clients based on the statistics of the batch normalization layers while preserving the specificity of each client with different local batch normalization.
- Semi-supervised deep extreme learning machine for Wi-Fi based localization95
The experimental results in real indoor environments show that the proposed Semi-supervised Deep Extreme Learning Machine (SDELM) not only outperforms other compared methods but also reduces the calibration effort with the help of unlabeled data.
- 90
- Effects of Dietary Coconut Oil as a Medium-chain Fatty Acid Source on Performance, Carcass Composition and Serum Lipids in Male Broilers89
Results indicated that replacement of 75% of the soybean oil in diets with coconut oil is the optimum level to reduce fat deposition and favorably affect lipid profiles without impairing performance in broilers.
- SELM: Semi-supervised ELM with application in sparse calibrated location estimation88
Comparative experiments show that with same number of labeled samples, the proposed novel semi-supervised learning method SELM outperforms original ELM and back propagation (BP) network, especially in the case that the calibration data is very sparse.
- Cross-Dataset Activity Recognition via Adaptive Spatial-Temporal Transfer Learning87
An Adaptive Spatial-Temporal Transfer Learning (ASTTL) approach to tackle both of the above two challenges in cross-dataset HAR and can be used for both source domain selection and accurate activity transfer.
- Cross-position activity recognition with stratified transfer learning83
A Stratified Transfer Learning (STL) framework to perform both source domain selection and activity transfer and is based on the proposed Stratified distance to capture the local property of domains.
- 83
- Self-supervised Learning for Electroencephalogram: A Systematic Survey78
This article concentrates on integrating SSL frameworks with temporal EEG signals to achieve efficient representations and proposes a systematic survey of the SSL for EEG signals.
- Dual near-infrared fluorescence-based lateral flow immunosensor for the detection of zearalenone and deoxynivalenol in maize78
A novel dual near-infrared fluorescence-based lateral flow immunosensor developed can serve as an accurate and efficient approach for monitoring mycotoxins in agricultural products.
- 77
- A vision-based sign language recognition system using tied-mixture density HMM76
Experimental results show that the proposed methods can work well for the medium vocabulary SLR in the environment without special constraints and the recognition accuracy is up to 92.5%.
- Generative AI for Synthetic Data Generation: Methods, Challenges and the Future73
Advanced technologies that leverage these gigantic LLMs for the generation of task-specific training data are delved into, including methodologies, evaluation techniques, and practical applications, and the current limitations are discussed.
- Deep unsupervised multi-modal fusion network for detecting driver distraction73
Experimental results show that UMMFN has superior performance over the existing approaches for distracted driving behavior, and empirically compare the proposed approach with several state-of-the-art methods on the own multi-modal dataset for distracteddriving behavior.
- 73
- Cross-domain activity recognition via substructural optimal transport68
This paper proposes substructure-level matching for domain adaptation (SSDA) to better utilize the locality information of activity data for accurate and efficient knowledge transfer and proposes an optimal transport-based implementation, Substructural Optimal Transport (SOT), for cross-domain HAR.
- Out-of-Distribution Representation Learning for Time Series Classification67
This paper proposes DIVERSIFY to learn generalized representations for time series classification through an iterative process that first obtains the worst-case distribution scenario via adversarial training, then matches the distributions of the obtained sub-domains.
- Development of an Immunochromatography Strip for the Rapid Detection of 12 Fluoroquinolones in Chicken Muscle and Liver67
A rapid and sensitive colloidal gold immunochromatography test strip based on one monoclonal antibody with broad-specificity, which can detect 12 fluoroquinolones (FQs), was developed and can be potentially used as a screening tool for the determination of 12 FQ residues in a large amount of samples on site.
- Domain Generalization for Activity Recognition via Adaptive Feature Fusion66
This article proposes Adaptive Feature Fusion for Activity Recognition (AFFAR), a domain generalization approach that learns to fuse the domain-invariant and domain-specific representations to improve the model’s generalization performance.
- Ten Challenging Problems in Federated Foundation Models64
This paper provides a comprehensive summary of the ten challenging problems inherent in FedFMs, encompassing foundational theory, utilization of private data, utilization of private data, continual learning, unlearning, Non-IID and graph data, bidirectional knowledge transfer, incentive mechanism design, game mechanism design, model watermarking, and efficiency.
- A Class Incremental Extreme Learning Machine for Activity Recognition64
This work proposes a class incremental extreme learning machine (CIELM), which builds an activity recognition model from labeled samples using anextreme learning machine algorithm without iterations and only requires labeled samples of new activities and not previously used training data.
- BLEselect58
A compact antenna array that fits on a pair of smart glasses to estimate the Angle of Arrival (AoA) of IoT and wrist-worn devices' advertising signals and a sensing pipeline that supports all three selection gestures with lightweight machine learning models, which are trained in real-time for both hand gestures.
- Diversify: A General Framework for Time Series Out-of-Distribution Detection and Generalization56
Qualitative and quantitative results demonstrate that Diversify learns more generalized features and significantly outperforms other baselines, as well as theoretical insights support the framework's validity.
- Algorithmic Management for Improving Collective Productivity in Crowdsourcing54
The surprise-minimization-value-maximization (SMVM) approach is proposed, which established a simple and novel worker desirability index (WDI) jointly considering the effect of each worker’s reputation, workload and motivation to work on collective productivity.
- Efficient Task Sub-Delegation for Crowdsourcing54
By jointly considering a worker's reputation, workload, the price of its effort and its trust relationships with others, RTS can be implemented as an intelligent agent to help workers make sub-delegation decisions in a distributed manner.
- ContextSense: unobtrusive discovery of incremental social context using dynamic bluetooth data53
Experimental results show that ContextSense can automatically cope with "incremental social context" classes that appear unpredictably in the real-world, and an ELM-based learning method for continuous and unobtrusive discovery of new social contexts incrementally from dynamic bluetooth data.
- Metagenomic insight into the microbial degradation of organic compounds in fermented plant leaves52
This work used a metagenomics sequence-guided strategy to identify the microbes and enzymes involved in compound degradation and explore the potential synergy among community members in fermented tobacco leaves, showing that contents of protein, starch, pectin, lignin, and cellulose varied in fermented leaves from different growing sites.
- TOSELM: Timeliness Online Sequential Extreme Learning Machine52
Experimental results show that TOSELM utilizing adaptive weight scheme and iteration scheme can achieve higher learning accuracy, faster convergence and better stability than other machine learning methods.
- Joint Source-Channel Coding and Optimization for Layered Video Broadcasting to Heterogeneous Devices52
- 51
- Mapping learning in eigenspace for harmonious caricature generation50
Experiments proved comparatively that the proposed mapping learning approach for caricature auto-generation could generate more harmonious caricatures.
- Wearing-independent hand gesture recognition method based on EMG armband47
To eliminate the influence of wearing position, Standard Space is proposed in this paper and, with the unified signals, fine hand gesture can be recognized accurately and robustly with lightweight Random Forest (RF).
- Extreme learning machine-based device displacement free activity recognition model47
A fast, robust and device displacement free activity recognition model, which integrates principal component analysis (PCA) and extreme learning machine (ELM) to realize location-adaptive activity recognition is proposed.
- Surface enhanced Raman scattering-based lateral flow immunosensor for sensitive detection of aflatoxin M1 in urine45
The developed immunosensor is a promising tool for the rapid assessment of aflatoxin exposure in the field and can achieve as low as 1.7 pg/mL of AFM1 in urine, which is far below the recommended tolerable level (30 PG/mL).
- Iterative Self-Training Based Domain Adaptation for Cross-User sEMG Gesture Recognition43
An Iterative Self-Training based Domain Adaptation method (STDA) is proposed to supervise the feature decoupling process with the pseudo-label generated by self-training and to explore cross-user sEMG gesture recognition.
- Deep and low-level feature based attribute learning for person re-identification42
A CNN-based pedestrian attribute-assisted person re-identification framework that performs the attribute learning by a part-specific CNN to model attribute patterns related to different body parts and fuse them with low-level robust Local Maximal Occurrence features to address the problem of the large variation of visual appearance and location of attributes.
- 42
- FallAlarm: Smart Phone Based Fall Detecting and Positioning System42
The FallAlarm system, utilizing a tri-axial accelerometer sensor and a Wi-Fi module embeded in the phone, can achieve an acceptable accuracy for practical application.
- 42
- 41
- Generalizable Low-Resource Activity Recognition with Diverse and Discriminative Representation Learning40
This paper proposes a novel approach called Diverse and Discriminative representation Learning (DDLearn) for generalizable low-resource HAR, which simultaneously considers diversity and discrimination learning and enhances semantic discrimination by learning discriminative representations with supervised contrastive learning.
- 40
- Feature Adaptive Online Sequential Extreme Learning Machine for lifelong indoor localization40
A Feature Adaptive Online Sequential Extreme Learning Machine (FA-OSELM) algorithm that can transfer the original model to a new one with a small number of data with new features, so as to make the new model suitable for the new feature dimension.
- Custom prosthetic reconstruction for proximal tibial osteosarcoma with proximal tibiofibular joint involved40
Custom prosthetic reconstruction could yield as much satisfactory results as neoadjuvant chemotherapy and limb salvage surgery, for patients with osteosarcoma in proximal tibia and proximalTibiofibular joint.
- 39
- 37
- Learning Critically: Selective Self-Distillation in Federated Learning on Non-IID Data36
This paper proposes a Selective Self-Distillation method for Federated learning (FedSSD), which imposes adaptive constraints on the local updates by self-distilling the global model's knowledge and selectively weighting it by evaluating the credibility at both the class and sample level.
- 36
- 35
- Inferring social contextual behavior from bluetooth traces35
Experimental results indicate that the proposed feature based classification approach and context restraint based majority vote rule can achieve high accuracy for inferring social contextual behavior through the real-life Bluetooth traces.
- Accurate and low-cost location estimation using kernels35
A novel approach to building a mapping between the signalvector space and the physical location space using kernel canonical correlation analysis (KCCA), which requires much less human labor than previous methods.
- ThermalRing: Gesture and Tag Inputs Enabled by a Thermal Imaging Smart Ring34
ThermalRing, a thermal imaging smart ring using low-resolution thermal camera for identity-anonymous, illumination-invariant, and power-efficient sensing of both dynamic and static gestures, and ThermalTag, thin and passive thermal imageable tags that reflect the heat from the human hand are proposed.
- b-COELM: A fast, lightweight and accurate activity recognition model for mini-wearable devices34
Experimental results show the efficiency and effectiveness of b-COELM for recognizing activities of daily living and extends the powerful generalization ability of Extreme Learning Machine in multi-class classification problems.
- 34
- Development of an enzyme-linked immunoassay for the detection of gentamicin in swine tissues34
An enzyme-linked immunoassay that provides rapid and sensitive detection of gentamicin in swine tissues and it showed negligible cross-reactivity with other aminoglycosides is developed.
- CLC: A Consensus-based Label Correction Approach in Federated Learning33
A Consensus-based Label Correction approach (CLC) in FL, which tries to correct the noisy labels using the developed consensus method among the FL participants, and proves the advantage over the state-of-art methods.
- 33
- Highly sensitive SERS immunosensor for the detection of amantadine in chicken based on flower-like gold nanoparticles and magnetic bead separation32
A novel ultrasensitive surface-enhanced Raman scattering (SERS) immunosensor based on the flower-like gold nanoparticles and magnetic bead separation for homogeneous detection of amantadine (AMD) in chicken just by one-step is reported.
- Duplex Surface Enhanced Raman Scattering-Based Lateral Flow Immunosensor for the Low-Level Detection of Antibiotic Residues in Milk31
The results demonstrate that the SERS-based lateral flow immunosensor developed in this study has the advantages of excellent assay sensitivity and remarkable multiplexing capability, thus it will have great application potential in food safety monitoring.
- Eating detection and chews counting through sensing mastication muscle contraction31
This paper presents the first effort in using motion sensor to sense mastication muscle contraction for continuous dietary monitoring through attaching a triaxial accelerometer on the temporalis and the results are promising.
- Semi‐Supervised Learning in Reconstructed Manifold Space for 3D Caricature Generation31
The experiments show that the novel approach synthesizes the 3D caricature more effectively than traditional methods, and has been applied successfully in a massive multi‐user educational game to provide human‐like avatars.
- FedEYE: A scalable and flexible end-to-end federated learning platform for ophthalmology29
FedEYE, a scalable and flexible end-to-end ophthalmic federated learning platform, is designed and developed, ensuring that ophthalmologists can effortlessly create independent and federated AI research tasks.
- FSELM: fusion semi-supervised extreme learning machine for indoor localization with Wi-Fi and Bluetooth fingerprints29
Fusion semi-supervised extreme learning machine (FSELM) is introduced, a novel semi- supervised learning algorithm based on the fusion of information from Wi-Fi and Bluetooth Low Energy signals that yields good results when applied to a multi-signal-based semi- Supervised learning problem.
- OKRELM: online kernelized and regularized extreme learning machine for wearable-based activity recognition29
OKRELM is a lightweight incremental learning model with less time consumption during the update and prediction phase, a robust and effective classifier compared with the batch learning scheme, and the performance is evaluated and compared with several related approaches on a UCI online available AR dataset.
- WeDA: Designing and Evaluating A Scale-driven Wearable Diagnostic Assessment System for Children with ADHD28
This study iteratively designed and tested WeDA, a scale-driven wearable diagnostic assessment system, and implements ten diagnostic tasks drawing on the symptoms of ADHD based on DSM-5, to address one specific concern about the objectivity and quantification of ADHD diagnosis.
- Accelerating Deep Unsupervised Domain Adaptation with Transfer Channel Pruning28
Experimental results demonstrate that TCP achieves comparable or better classification accuracy than other comparison methods while significantly reducing the computational cost.
- 27
- Learning and synthesizing MPEG-4 compatible 3-D face animation from video sequence27
A new system that applies an example-based learning method to learn facial motion patterns from a video sequence of individual facial behavior, and using that to create vivid three-dimensional (3-D) face animation according to the definition of MPEG-4 face animation parameters.
- 27
- Effects of organic chromium sources on growth performance, lipid metabolism, antioxidant status, breast amino acid and fatty acid profiles in broilers26
Diets supplemented with organic Cr could regulate lipid metabolism, and improve amino acid and fatty acid profiles in broiler breast, whereas Cr‐yeast was the most effective source in improving methionine and cysteine content and CrMet was more effective than CrNic in increasing n‐3 PUFA value and decreasing n‐6/n‐3PUFA ratio in breast meat.
- CARE: chewing activity recognition using noninvasive single axis accelerometer26
A novel recognition system named CARE is presented that uses a single axis accelerometer attached on the temporalis to detect the muscle bulge and recognize user's chewing activity and is able to calculate the chewing frequency through recognizing the periodic pattern of acceleration data.
- GestureGPT: Toward Zero-Shot Free-Form Hand Gesture Understanding with Large Language Model Agents25
G gestureGPT is introduced, a free-form hand gesture understanding framework that mimics human gesture understanding procedures to enable a natural free-form gestural interface and validated offline under two real-world scenarios: smart home control and online video streaming.
- Dual layer transfer learning for sEMG-based user-independent gesture recognition25
This work proposes a dual layer transfer learning framework, named dualTL, to realize user-independent gesture recognition based on sEMG signal, and shows that the average recognition accuracy of dualTL is 80.17%.
- 25
- Ultra-High Resolution Segmentation via Boundary-Enhanced Patch-Merging Transformer24
Inspired by the human visual system's ability to rapidly orient attention to important areas with fine details and filter out irrelevant information, a novel UHR segmentation method called Boundary-enhanced Patch-merging Transformer (BPT) is proposed that outperforms previous state-of-the-art methods without introducing extra computational overhead.
- 24
- A Feature Adaptive Learning Method for High-Density sEMG-Based Gesture Recognition24
A random forest based ensemble learning method, namely feature incremental and decremental ensemble learning (FIDE), which significantly improves the gesture recognition accuracy in electrode increment and decrement circumstances and implements a multi clients/server collaboration system, namely McS, to support feature adaption in real-world environment.
- Lateral flow immunoassay for 5-hydroxyflunixin based on near-infrared fluorescence molecule as an alternative label to gold nanoparticles23
Results demonstrated that the NIR dye is an ideal alternative label to the conventional AuNP for the development of LFA for veterinary drugs in animal-origin food.
- 23
- 23
- 22
- 22
- Pedestrian Attribute Recognition with Part-based CNN and Combined Feature Representations22
A neural network combining high-level learnt Convolutional Neural Network features and low-level handcrafted features to address the problem of highly varying viewpoints and fuse the two feature representations to recognise pedestrian attributes.
- 22
- 22
- FIXED: Frustratingly Easy Domain Generalization with Mixup21
This work proposes a simple yet effective enhancement for Mixup-based DG, namely domain-invariant Feature mIXup (FIX), which learns domain- Invariant representations for Mix up and significantly outperforms nine state-of-the-art related methods.
- What can “drag & drop” tell? Detecting mild cognitive impairment by hand motor function assessment under dual-task paradigm21
Cortical activation analysis shows that bilateral prefrontal cortices, bilateral motor cortices and the occipital lobe are involved in “drag & drop” tasks, proving the effectiveness of cogSYS in specific cognitive functions.
- 21
- Surrounding context and episode awareness using dynamic Bluetooth data21
This paper extracts novel features from Bluetooth traces and uses these features for recognizing contextual behavior as well as inferring continuous episode transition and achieves an average of 87% accuracy for specific context classification and the ability of episode inference from real-life Bluetooth traces.
- Recognition of sign language subwords based on boosted hidden Markov models21
Experimental results on the vocabulary of frequently used Chinese sign language (CSL) subwords show that the proposed boosted CHMM outperforms the conventional CHMM for SLR.
- GJFusion: A Channel-Level Correlation Construction Method for Multimodal Physiological Signal Fusion20
The results demonstrate that the proposed GJFusion network surpasses the latest state-of-the-art methods, achieving relative accuracy improvements of 1.22% and 0.81% on the DEAP and MAHNOB-HCI datasets, respectively.
- Monitoring perception reliability in autonomous driving: Distributional shift detection for estimating the impact of input data on prediction accuracy20
This research evaluates state of the art OOD detection methods on autonomous driving camera data, while also demonstrating the influence of OOD data on the prediction reliability of neural networks.
- A Mobile Cloud Collaboration Fall Detection System Based on Ensemble Learning20
Experiments show that the proposed FEDT outperforms the others' over 1-3% both on sensitivity and specificity and has superior robustness on different devices.
- 20
- Semi-supervised Learning of Caricature Pattern from Manifold Regularization20
A novel pattern prediction method based on MR (manifold regularization), which comprises three steps, where low dimensional manifold represents the caricature pattern well and the semi-supervised regressive model from manifold regularization can predict the target caricature pattern successfully.
- Preparation of Acetochlor Antibody and Its Application on Immunoaffinity Chromatography Cleanup for Residue Determination in Peanuts20
The preparation of an antibody against acetochlor and its application on immunoaffinity chromatography (IAC) cleanup was investigated and the recoveries from spiked peanut samples ranged from 94.7 to 102.5%, with a variation coefficient of 2.6-8.0%.
- 19
- 19
- 19
- A ring-shaped interactive device for large remote display and mobile device control19
A novel human-computer interaction device is proposed to realize finger touching for large display and mobile device control, without a touchscreen or a touch pad, using bone-conducted friction sound, acceleration and gyroscope sensor data.
- Cross-mobile ELM based Activity Recognition19
This work proposes an algorithm known as TransELMAR(Transfer learning and Extreme Learning Machine based Activity Recognition) that integrates the transfer learning technique and extreme learning machine algorithm for activity recognition model adaptation that outperforms several traditional baseline algorithm.
- 18
- FewShotBP18
A novel few-shot transfer learning approach named FewShotBP is proposed, which addresses the above-mentioned challenges by introducing a personalization adapter at the personalization stage, and a multi-modal spectro-temporal neural network at the pre-train stage, to bridge the gap between data-hungry models and limited labeled data in realistic scenarios.
- Class-Wise Adaptive Self Distillation for Federated Learning on Non-IID Data (Student Abstract)18
This paper designs class-wise adaptive terms to soften the influence of distillation loss according to the global model's performance on each class and therefore avoid the misleading of local models.
- A dataset of human decision-making in teamwork management18
This paper describes a multi-faceted large-scale dataset derived from a game simulating complex project management processes that contains detailed data reflecting the decision situations, decision strategies, decision outcomes, and the emotional responses of 1,144 participants from diverse backgrounds.
- AI empowered context-aware smart system for medication adherence18
An AI-empowered context-aware smart pillbox system that uses argumentation-based activity recognition and reminder generation that will generate reminders at appropriate time and on appropriate devices is presented.
- MINING USER GOALS FOR INDOOR LOCATION‐BASED SERVICES WITH LOW ENERGY AND HIGH QoS18
This article designs new algorithms to mine user goals from discontinuous location data to reduce the energy consumption while keeping a high QoS at the same time, and shows that the QoS of LBS can be evaluated by recognizing user goals.
- Intraoperative blood pressure and cardiac complications after aneurysmal subarachnoid hemorrhage: a retrospective cohort study17
Lower blood pressures were associated with cardiovascular complications over the entire observed range, but worsened when MAP was less than 75 mmHg, and it may be prudent to keep MAP above 75 mmHg in patients having surgical aSAH repairs to reduce the risk of MACE.
- 17
- An improved algorithm for personalized recommendation on MOOCs17
A multi-attribute weight algorithm based on collaborative filtering (CF) to select a recommendation set of courses for target MOOC users by reflecting the target users’ preferences for the first time by calculating separately the weight of the attributes and the Weight of attribute values of the courses.
- 17
- 17
- Gravity-Center Template Based Human Face Feature Detection17
This paper presents a simple and fast technique for geometrical feature detection of several human face organs such as eyes and mouth by using human face gravity-center template for face location.
- 16
- 16
- 16
- 15
- Multi-Layer Cross Loss Model for Zero-Shot Human Activity Recognition15
A novel model termed Multi-Layer Cross Loss Model (MLCLM), which designs a multi-nonlinear layers model to project features to semantic space for that the deeper the network is, the better the network can fit the data's distribution.
- Multi-Agent System Development MADE Easy15
The MADE system is outlined, which provides an interactive platform for people who are not well-versed in AOSE to contribute to the rapid prototyping of MASs with ease.
- A semi-supervised privacy-preserving clustering algorithm for healthcare15
Experimental results on the brain fiber dataset provided by the 2009 PBC demonstrated that the proposed semi-supervised privacy-preserving clustering algorithm could not only protect data privacy towards secure attacks, but improve the clustering purity.
- Interactive 3D caricature generation based on double sampling15
The experiments show that the 3D caricature generated by the proposed novel interactive approach can preserve highly artistic styles and also reflect the user's intention.
- Exploring Structure Incentive Domain Adversarial Learning for Generalizable Sleep Stage Classification14
This work proposes a Structure Incentive Domain Adversarial learning (SIDA) method that combines the sleep stage classification method with domain generalization to enable cross-subject sleep stage Classification and demonstrates that the proposed SIDA method outperforms other state-of-the-art sleep stage classified methods and achieves the best cross- subjectSleep stage classification results.
- Two-step hierarchical neural network for classification of dry age-related macular degeneration using optical coherence tomography images14
This study introduces a novel two-step hierarchical deep learning approach in categorizing dry AMD progression phases, and demonstrates its efficacy, and suggests its potential for guiding individualized treatment plans for patients with macular degeneration.
- 14
- Design of Hybrid Control Algorithm for Fault Ride-Through of Photovoltaic System14
It can be proved that the major problems troubling a PV system’s FRT technologies can be resolved by the proposed method, especially the problem related to the harmonic components of the DC-link voltage.
- Low voltage ride-through capability improvement of microgrid using a hybrid coordination control strategy14
The related theoretical derivation and simulation analysis confirm that the proposed control strategy can not only enhance the LVRT capability, but also strengthen the microgrid transient performance.
- 14
- ProCom: designing a mobile and wearable system to support proximity awareness for people with autism14
A prototype system, ProCom, is developed that combines wearable sensors with a mobile phone application to support developing awareness of physical proximity in social settings without requiring instrumentation of the room or other person.
- RECOGNIZING TRANSPORTATION MODE ON MOBILE PHONE USING PROBABILITY FUSION OF EXTREME LEARNING MACHINES14
A transportation mode recognition method using probability fusion of extreme learning machines (ELMs) that can obtain higher recognition accuracy than traditional fusion methods.
- 14
- An Activity Transition Based Fall Detection Model on Mobile Devices14
A fall detection model based on activity transition is proposed that can detect the fall through recognizing abnormal activity transition and outperforms traditional methods on the real-world activity data set.
- 14
- Immunoassays for the Rapid Detection of Gentamicin and Micronomicin in Swine Muscle14
The detection results of incurred samples analyzed by the strip test, ELISA, and HPLC indicated that the two immunoassays correlated well with the HPLC method and could be used as convenient tools for the rapid screening of GEN and MIN residues in swine muscle.
- 13
- 13
- 13
- Instance-Wise Dynamic Sensor Selection for Human Activity Recognition13
Experimental results show that IDSS can effectively reduce the overall sensor number without losing accuracy and outperforms the state-of-the-art methods regarding the combined measurement of accuracy and sensor number.
- Transfer channel pruning for compressing deep domain adaptation models13
Experimental results demonstrate that the proposed unified transfer channel pruning (TCP) method achieves comparable or better classification accuracy than other comparison methods while significantly reducing the computational cost.
- BrainStorm: a psychosocial game suite design for non-invasive cross-generational cognitive capabilities data collection13
An extensive evaluation of the presented psychosocial game suite demonstrated high acceptability and attraction for its target users and the data collection process is successfully reported as transparent and non-invasive.
- 12
- Bi-STAN: bilinear spatial-temporal attention network for wearable human activity recognition12
An efficient Bilinear Spatial-Temporal Attention Network (Bi-STAN), where a multi-scale ResNet backbone network is used to extract multimodal signal features and jointly optimize the feature extraction process, and a bilinear pooling with low redundancy is introduced to efficiently obtain second-order information.
- 12
- 12
- ASELM: Adaptive semi-supervised ELM with application in question subjectivity identification12
An adaptive semi-supervised Extreme Learning Machine (ASELM) is proposed to solve the data imbalance problem in Community Question Answering, and outperformed the performance of basic ELM, SELM, Weighted ELM and SS-ELM on both F1 measure and accuracy.
- OCEAN: a new opportunistic computing model for wearable activity recognition12
An Opportunistic Computing model for wEarable Activity recognitioN (OCEAN) is proposed, which can online adaptively adjust Single-hidden Layer Feedforward neural network's connection, structure and weight in a coherent manner by fusing random mapping, fuzzy clustering, and weight updating techniques.
- AIR: recognizing activity through IR-based distance sensing on feet12
The results show that AIR obtains much higher accuracy in recognizing activity than approaches that rely primarily on accelerometers and has good generalization ability when applying recognition model to new users.
- Leveraging Two-Stage Weighted ELM for Multimodal Wearables Based Fall Detection12
Experimental results show that the method is able to effectively implement on miniaturized wearable devices, and compared to state-of-the-art ELM classifier, it can also obtain higher detection accuracy and lower false-alarm rate simultaneously, which enables various kinds of mHealth applications in large-scale population.
- LEAPS: A Location Estimation and Action Prediction System In a Wireless LAN Environment12
An integrated framework called LEAPS (location estimation and action prediction), jointly developed by Hong Kong University of Science and Technology and the Institute of Computing, Shanghai, is presented that combines two areas of interest, namely, location estimation and plan recognition, in a coherent whole.
- LEAPS: A Location Estimation and Action Prediction System in a Wireless LAN Environment12
An integrated framework called LEAPS (location estimation and action prediction), jointly developed by Hong Kong University of Science and Technology and the Institute of Computing, Shanghai, is presented that combines two areas of interest, namely, location estimation and plan recognition, in a coherent whole.
- Biotechnology revival: in situ sludge minimization in wastewater11
This review comprehensively highlights the fundamental mechanisms, application strategies, and future prospects of in situ sludge reduction biotechnology, including one of the first studies to introduce bacteriophage-based approaches for in situ sludge reduction, offering a novel perspective on phage-mediated sludge control.
- Correlation-Driven Multi-Modality Graph Decomposition for Cross-Subject Emotion Recognition11
This study introduces a novel framework, termed Correlation-Driven Multi-Modality Graph Decomposition (CMMGD), which captures adaptive cross-modal correlations and incorporates a correlation-driven graph decomposition module that decomposes the mixed graph into concordant and discrepant subgraphs based on the correlations.
- 11
- 11
- 11
- Enzyme immunoassay and liquid chromatography-fluorescence detection for amikacin in raw milk11
An enzyme immunoassay and a liquid chromatography method for amikacin (AMK) inRaw milk were developed in this study and can be selectively used for rapidly screening or quantitatively determining AMK in raw milk.
- The Harris Corner Detection Method Based on Three Scale Invariance Spaces11
An improved three scale Harris corner detection algorithm was proposed, which proved that, compared with the scale space on the whole Gaussian pyramid, the utilization factor of the image was increased, the calculation time is decreased, and theimage was high recurrence rate and stability.
- Complete genome sequence of Bacillus cereus Z4, a biocontrol agent against tobacco black shank, isolated from the Western Pacific Ocean10
The whole genomic analysis of Bacillus cereus Z4 may provide a valuable reference for elucidating its biocontrol mechanism against tobacco black shank.
- 10
- Accurate and Robust Moving-Object Segmentation for Telepresence Systems10
This work proposes an adaptive multi-cue decision fusion method based on Kinect (which integrates a depth sensor with an RGB camera) that can segment moving objects accurately and robustly in various situations in real time.
- 10
- 10
- Learning Prosodic Patterns for Mandarin Speech Synthesis10
A combination of clustering and machine learning techniques are proposed to extract prosodic patterns from actual large mandarin speech databases to improve the naturalness and intelligibility of synthesized speech.
- An adaptive joint optimization framework for pruning and quantization9
This work proposes A-JOPQ (adaptive joint optimization of pruning and quantization), an adaptive joint optimization framework for pruning and quantization that achieves higher accuracy with a smaller model size and can enhance federated learning (FL) settings.
- 9
- Two-stage deep neural network for diagnosing fungal keratitis via in vivo confocal microscopy images9
A deep learning framework for diagnosing fungal keratitis using IVCM images to assist ophthalmologists using a two-stage deep architecture for diagnostic predictions based on both image-level and sequence-level information.
- 9
- 9
- 9
- ESVD: An Integrated Energy Scalable Framework for Low-Power Video Decoding Systems9
This paper is dedicated to developing an energy-scalable video decoding (ESVD) strategy for energy-limited mobile terminals that combines the decoder control with decoded data, and introduces utility theoretical analysis during the resource allocation process, so as to maximize the resource utilization.
- Automatic generation of pencil sketch for 2D images9
The experimental results prove that the novel method put forward could efficiently generate more obvious edge, more natural tone and more real texture than existed methods.
- Transfer Regression Model for Indoor 3D Location Estimation9
The experimental results show that by TRM, the approach of indoor mobile 3D location estimation based on TRM can achieve higher accuracy with sparse training dataset compared to the regression model without knowledge transfer.
- 9
- 9
- FedIERF: Federated Incremental Extremely Random Forest for Wearable Health Monitoring8
A horizontal federated learning method called Federated Extremely Random Forest (FedERF), which introduces a hardness-driven weighting mechanism and an importance-based updating scheme to update the existing federated model incrementally and effectively addresses the dynamic expansion of the federation.
- 8
- 8
- Quality-time-complexity universal intelligence measurement8
A calculable universal intelligent measure method through considering more than two factors and the correlations between factors which are involved in an intelligent measurement is proposed.
- 8
- Cooperative mobile video transmission for traffic surveillance in smart cities8
This study proposes a cooperative mechanism for mobiles that can make full use of local storage space and uploading bandwidth to relieve the computational load for streaming servers.
- 8
- Real-Time and Accurate Indoor Localization with Fusion Model of Wi-Fi Fingerprint and Motion Particle Filter8
In this framework, Extreme Learning Machine (ELM) regression algorithm is used to predict position based on motion sensors and use Wi-Fi fingerprint location result to solve the error accumulation of motion sensors based location occasionally with Particle Filter.
- 8
- 8
- A perceptual macroblock layer power control for energy scalable video encoder based on just noticeable distortion principle8
Experimental results demonstrate the efficiency of the proposed Macroblock (MB) level perceptual energy scalable video encoding method, PMP-ESVE, in which just noticeable distortion (JND) model is introduced as the perceptual cue.
- Monoclonal Antibody-Based Immunoassay for the Detection of Maduramicin in Chicken Tissues8
The Mab with high anti-interference ability and high affinity was selected, and it exhibited negligible cross-reactivity with other usual-used polyether antibiotics and showed an IC50 value of 2.12 ± 0.46 ng/ml after optimization.
- Development of a Monoclonal Antibody-Based ELISA for the Detection of Sulfaquinoxaline in Chicken Tissues and Serum8
The anti-SQX Mab exhibited negligible cross reactivity with other commonly used sulfonamides, and a competitive indirect enzyme-linked immunosorbent assay (ciELISA) was developed to detect SQX in chicken tissues and serum.
- 7
- 7
- Human-centered intelligent healthcare: explore how to apply AI to assess cognitive health7
Inspired by clinically validated cognition assessment test—Trail Making Test (TMT), two variants of TMT are designed for objective quantitative assessment, including camera-based TMT (cTMT) and touchscreen-basedTMT (tT MT), which can achieve satisfactory discrimination accuracy by optimizing the assessment model.
- 7
- 7
- Precise locating approach of the beacon based on gray gradient segmentation interpolation in satellite optical communications7
Two precise centroid computation methods for locating a beacon in satellite optical communications are presented and results indicate that GSI and SSW algorithms can improve locating accuracy over that calculated by a traditional gray centroid method.
- Numerical simulation based on three-dimensional model of inner stereocilia7
The results show that the movement of the hair cells affects the x-displacements of the inner stereocilia, especially for the shortest stereocilium, and the fluid in the cochlear duct affects thex-dis Placement of the Inner stereOCilia,especially for the middle stereOCilium.
- A Study of Players’ Experiences During Brain Games Play7
An empirical study that assesses the experience of 50 participants during brain games play shows a number of significant correlations among diverse kinds of players' experiences and exploits these insights to narrow down the complexity of user feedback process for brain games playing activity.
- Indoor Location Estimation Based on Local Magnetic Field via Hybrid Learning7
In this chapter, the magnetic field samples over more than 2 months inside an office building presents a finding that there exist the relative stable measurements for a single location and the relative obvious difference in most of locations and a hybrid learning method based on the local magnetic field measurements is proposed.
- 7
- Personalized 3D Caricature Generation Based on PCA Subspace7
The approach is very convenient and efficient to generate different kinds of 3D caricature by personalized thought, for example, a user just needs to drag the sliders for each principal component of the3D caricature component.
- Content-Based Human Motion Retrieval with Automatic Transition7
A Scene Description Language that can translate user's input into a series of set operations between inverted lists and has three-layer structures, each describing scenes at different levels of granularity is proposed.
- Motion Retargeting for the Hand Gesture7
This paper presents a new technique for retargeting the sign language data captured from motion capture device to different characters with different sizes and proportions and dramatically improves the recognition rate.
- Unified model in identity subspace for face recognition7
A unified model in identity subspace (UMIS) as a novel technique for face recognition from only one example image per person, spanned by bilateral symmetric bases, which compactly encodes identity information.
- 6
- 6
- SleepMG: Multimodal Generalizable Sleep Staging with Inter-modal Balance of Classification and Domain Discrimination6
Experimental results on public sleep staging datasets demonstrate that SleepMG outperforms state-of-the-art sleep staging methods, effectively balancing multiple modalities as evidenced by the visual experiment of modal imbalance degree.
- 6
- 6
- List-wise learning-to-rank with convolutional neural networks for person re-identification6
The approach is evaluated on two different computer vision applications that are commonly formulated as ranking problems: person re-identification and image retrieval with several public benchmarks and showed that the new loss function outperforms other common functions and that the method achieves state-of-the-art performance compared to existing approaches from the literature.
- 6
- Bridging Cross-Tasks Gap for Cognitive Assessment via Fine-Grained Domain Adaptation6
This paper proposes a novel domain adaptation method, namely the Fine-Grained Adaptation Random Forest (FAT), to bridge the cognitive assessment gap when the data distribution is changed.
- A Lightweight Fully Convolutional Neural Network of High Accuracy Surface Defect Detection6
A lightweight fully convolutional neural network, named LFCSDD is proposed, which can process the images with different sizes, and obtain the accuracy of 99.72% and 98.74% on benchmark defect datasets, DAGM 2007 and KolektorSDD, respectively, outperforming all the baselines.
- 6
- 6
- 6
- 6
- Rate-Distortion-Complexity Analysis on AVS Encoder6
By adjusting the modules and parameters according to the analysis, the power or complexity of the encoding device can be efficiently reduced to nearly 10% with only small changes of RD cost, and power-scalable or complexity-configurable encoder can be further designed.
- 6
- A new all-zero 4×4 block determination rule for integer transform and quantization in AVS-M encoder6
A new all-zero block determination rule was used to reduce the complexity of the AVS-M encoder and can save about 20%, which avoids a lot of computation for transform, quantization, inverse transform, inverse quantization and block reconstruction.
- 6
- 6
- Multi-strategy data mining on Mandarin prosodic patterns6
A Multi-strategy Data Mining framework to extract prosodic patterns from actual largeMandarin speech database to improve the naturalness and intelligibility of synthesized speech and showed that synthesized prosodic features quite resembled their original counterparts for most syllables.
- Mitigating Pervasive Modality Absence Through Multimodal Generalization and Refinement5
Comprehensive experiments on four benchmark datasets demonstrate that the proposed MGR framework outperforms state-of-the-art methods, effectively mitigating the impact of pervasive modality absence.
- 5
- Buffalo: Biomedical Vision-Language Understanding with Cross-Modal Prototype and Federated Foundation Model Collaboration5
The results demonstrate that Buffalo can fully utilize data from all clients to improve model generalization compared to other modal imputation methods in three modal heterogeneity scenarios, approaching or even surpassing the performance in the ideal scenario without missing modality.
- 5
- WebJump: AR-facilitated Distributed Display of Web Pages5
This paper proposes WebJump, a software tool that analyzes HTML web pages and enables UI elements ‘jump’ from the PC monitor to the HMD, so that contents like high-resolution images and long texts can take full advantage of the high-quality monitor display.
- 5
- KiCi5
A knowledge importance-based class incremental learning method called KiCi is proposed and constructed based on the framework of self-iterative knowledge distillation for dynamic activity recognition and consistently outperforms the existing state-of-the-art methods by a large margin.
- 5
- 5
- Two-Layer Optimization Model for the Siting and Sizing of Energy Storage Systems in Distribution Networks5
A two-layer optimization model to determine the optimal siting and sizing of ESSs in the distribution network and their best compromise between the real power loss, voltage stability margin, and the application cost is presented.
- 5
- 5
- Anomaly data management and big data analytics: an application on disability datasets5
This paper proposed a big data management and analytic approach for disability datasets that can improve the quality of disability data by estimating miss attribute values and detecting anomaly and low-quality data instances.
- Less annotation on active learning using confidence-weighted predictions5
An efficient and effective active online sequential learning approach that leverages the predictions’ confidence of the new arriving data to actively select both query-annotated samples and confidence-weighted predict-annotation ones to update the classifier, and applies WOS-ELM to significantly reduce the computation complexity for doing online updating in one step.
- Mechanical model of round window membrane under reverse excitation5
The analytical analysis can avoid interference of environment and the technical level of personnel, and it can evaluate transmission performance of the round window membrane efficiently, providing a theoretical basis for the reverse excitation of artificial prosthesis.
- Cross-Layer Opportunistic Scheduling for Device-to-Device Video Multicast Services5
This article proposes an opportunistic transmission and fair resource allocation framework, including joint application-layer and physical-layer transmission and optimization, and uses the utility fair scheme to guide rate allocation among multicontent video multicast.
- 5
- Online deep intelligence for Wi-Fi indoor localization5
A novel online deep learning approach OSDELM is proposed, which guarantees the running time of localization system from two aspects: discriminative feature, and updated model, and online learning updates the out-of-date model to fit the new environment.
- 5
- 5
- 5
- Creative Cartoon Face Synthesis System for Mobile Entertainment5
A prototype system, which synthesizes entertainment-oriented cartoon face and translates text message to multimedia animation in mobile phone using TTVS method and improved modified Active Shape Model to remove the background and control more feature points on the face is presented.
- 5
- Creative Cartoon Face Synthesis System for Mobile Entertainment5
A prototype system, which synthesizes entertainment-oriented cartoon face and translates text message to multimedia animation in mobile phone using TTVS method and improved modified Active Shape Model to remove the background and control more feature points on the face is presented.
- 5
- 4
- 4
- 4
- Circulating Biomarkers Predict Immunotherapeutic Response in Hepatocellular Carcinoma Using a Machine Learning Method4
Circulating biomarkers are useful in predicting therapeutic outcomes and can help in making clinical decisions regarding the use of ICI therapy.
- 4
- AUGlasses: Continuous Action Unit based Facial Reconstruction with Low-power IMUs on Smart Glasses4
AUGlasses achieves unobtrusive low-power facial reconstruction by placing inertial measurement units against the temporal area on the face to capture the skin deformations, which are caused by facial muscle movements.
- 4
- CellPolaris: Decoding Cell Fate through Generalization Transfer Learning of Gene Regulatory Networks4
This study introduces CellPolaris, a computational system that leverages transfer learning algorithms to generate high-quality, cell-type-specific GRNs, and represents a significant advancement in deciphering the mechanisms of cell fate regulation, thereby enhancing the precision and efficiency ofcell fate manipulation at high resolution.
- 4
- 4
- Smart Decision-Support System for Pig Farming4
A smart decision-support system for pig farming is presented that adopts rail-based unmanned vehicles to capture pigsty images and conducts image stitching to avoid double-counting pigs, and uses image segmentation method to give precise masks for each pig.
- 4
- 4
- Person Re-Identification with a Body Orientation-Specific Convolutional Neural Network4
This paper proposes an Orientation-Specific Convolutional Neural Network (OSCNN) framework which jointly performs body orientation regression and extracts orientation-specific deep representations for person re-identification.
- 4
- 4
- 4
- 4
- 4
- 4
- Ti-Photograph: a tele-immersive photograph system for distributed parents and children4
A Tele-Immersive Photograph system (TI-Photograph), which allows children and their remote parents to take pictures together in a virtual space with interactive behaviors and introduces a user behavioral intention driven video composition method to adaptively merge the segmented videos of the users into a same shared background customized by natural gesture interaction.
- 4
- 4
- Vision-Based Sign Language Recognition Using Sign-Wise Tied Mixture HMM4
Experimental results based on 439 frequently used Chinese sign language (CSL) signs show that the proposed methods can work well for the medium vocabulary SLR in the unconstrained environment.
- Speech Driven MPEG-4 Based Face Animation via Neural Network4
The performance of the system shows that the proposed learning algorithm is suitable, which can greatly improve the realism of real time face animation during speech.
- K-Space Bispectrum Steganography for Robust Unlearnable Data3
KBS (K-Space Bispectrum Steganography), which embeds class-specific information into the magnitude and phase components of the Fourier domain while preserving visual fidelity under reconstruction constraints, is proposed, which outperforms state-of-the-art methods.
- 3
- 3
- Auto-Augmentation Contrastive Learning for Wearable-based Human Activity Recognition3
This work proposes an end-to-end auto-augmentation contrastive learning method for wearable-based HAR based on a Siamese network architecture that shares the parameters of the backbone and with a generator embedded to learn auto-augmentation.
- 3
- 3
- A Weighted Multi-attribute Method for Personalized Recommendation in MOOCs3
A personalized recommendation algorithm---multi-attribute weight algorithm (MAWA) based on collaborative filtering technology, which can reflect the users' preferences in both coarse granularity and fine granularity.
- Mobile Social Multimedia Analytics in the Big Data Era3
This special issue will present new theories and new application scenarios on mobile social multimedia computing, communication, and application, and survey the recent progress in this area, as well as discuss new applications and problems.
- 3
- Distributed Video Coding: an overview of basics, research issues and solutions3
An overview of DVC is provided, including the theoretical foundation: Slepian-Wolf Theorem and Wyner-Ziv Theorem, the framework foundation: Distributed Source Coding (DSC) and the relation between DSC and DVC, and unique technologies comparison to conventional compressed method, which highlight critical research issues.
- 3
- 3
- 3
- 3
- Power Scalable Video Encoding Strategy Based on Game Theory3
The proposed PSVE uses game theory to solve the tradeoff between encoding effect and power consumption and can make video encoder work under variable energy resource constraint while keep better performance.
- 3
- 3
- Early detection of Parkinson's disease: Machine learning-based prediction of UPDRS Part III scores in de novo patients using smartphone assessments2
Smartphone-based assessments effectively discriminate de novo PD patients from controls and monitor motor symptoms in prodromal and early PD patients are suggest that smartphone-based assessments can be useful tools for identifying early motor deficits in de novo PD patients and monitoring of PD motor symptoms from home.
- Information Retrieval Optimization for Non-Exemplar Class Incremental Learning2
This work proposes a new NECIL method based on the IB framework that employs a pseudo-prototype selection mechanism that selectively incorporates pseudo-samples into the learning process of new and old categories, thus enhancing the distinction between new and old categories and diminishing the mutual information between the input and intermediate features.
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- Agent-based Decision Support for Pain Management in Primary Care Settings2
An Agent-based Clinical Decision Support System is demonstrated to empower PCPs to leverage knowledge from pain specialists and learns a general-purpose representation space on patients, which automatically diagnoses pain, recommends therapy and medicine, and suggests a referral program to PCPs in their decision-making tasks.
- Knowledge discovery in sociological databases2
A comprehensive data management and data mining approach for GSS data sets which can explore knowledge in a fine-grained manner with minimum human interference is proposed.
- Fast media caching for geo-distributed data centers2
This paper proposes Fmc (fast media caching), a cost-aware and highly efficient caching algorithm for vCDN delivery over geo-distributed data centers, and proves that Fmc is k k − h + 1 competitive under the resource augmentation paradigm.
- Quality-Complexity-Task Universal Intelligence Measurement2
This work proposes a formalized and accurate intelligence measurement method named Quality-Complexity-Task (QCT) universal intelligence measurement, and uses a kind of intelligence test method to measure agent intelligence.
- Person Re-identification Using Group Context2
This paper proposes to use a deep convolutional neural networks (CNN) to extract group feature representations that are invariant to the relative displacements of individuals within a group, and uses this group feature representation to perform group association under non-overlapping cameras.
- Opportunistic Work-Rest Scheduling for Productive Aging2
OWRS establishes a framework which accounts for workers’ heterogeneity to enhance their experience and productivity and significantly outperforms three baseline scheduling approaches in terms of conserving worker effort while achieving superlinear collective productivity.
- Evaluate mobile video quality in hybrid spatial and temporal domain2
A new hybrid spatial and temporal distortion metric for evaluating video streaming quality with variable spatio-temporal resolution with good correlations with the subjective evaluations of the EPFL scale video database is proposed.
- Wi-Fi and Motion Sensors Based Indoor Localization Combining ELM and Particle Filter2
An effective combined model is presented, which uses ELM regression algorithm to predict position based on emotion sensor, and then combines the Wi-Fi location result to motion sensor based location result using particle filter to get higher accurate location result.
- 2
- Semi-heterogeneous Wireless Location Method Based on Hidden State Sorting2
This paper introduces Hidden Markov Model (HMM) and hidden state sorting approach, and designs the localization method including offline training phase and online localization phase, which can achieve 96.3% accuracy within an error distance of 1 meter.
- Distributed Adaptive Channel Assignment in WLANs2
The extensive NS2 simulation result shows that the new algorithm is converged quickly, reliable and stable, so effectively increases the network overall throughout.
- 2
- 2
- 2
- A complexity scalable decoder in an AVS video codec2
This paper aims at the tradeoff between complexity and video quality, and proposes a complexity scalable AVS video decoder, which can accurately reduce K% computational complexity with slight video quality degradation.
- 2
- Unequal Error Protection Based on Objective Video Evaluation Model2
A novel effective and reliable objective video evaluation model based on fuzzy synthetic judgment which highlights a comprehensive evaluation by taking account of multiple properties of the compressed video which affect the whole video sequence, such as quality, fluency and motion information.
- 2
- 1
- 1
- 1
- 1
- 1
- 1
- A Mutual Learning Framework for Pruned and Quantized Networks1
A mutual learning framework for pruned and quantized networks to better integrate the two sets of features and achieve complementary advantages, which is called feature augmentation.
- 1
- 1
- BoldMove: Enabling IoT Device Control on Ubiquitous Touch Interfaces by Semantic Mapping and Sequential Selection1
This work proposed the semantic-based IoT function filtering to improve control efficiency, then designed the sequential selection mechanism for interfaces with constrained input and output resources and validated that BoldMove only takes 3.25 seconds to complete a selection task if the target function appears within the Top-3 displayed item.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- Big Data Management and Analytics for Disability Datasets1
The proposed approach can improve the quality of disability data by estimating miss attribute values and detecting anomaly and low-quality data instances, and explore useful patterns which reflect the correlation, association and interactional between the disability data attributes.
- 1
- 1
- 1
- 1
- 1
- QoE-based opportunistic transmission for video broadcasting in heterogeneous circumstance1
The main ideas proposed here include exploiting the quality of heterogeneous user experience (QoE) metric under wireless broadcasting scenario, with consideration of various channel state, device capability, video content urgency and the number of demanding users.
- 1
- AVS Encoder Performance and Complexity Co-evaluation Based on Mobile Video Communication1
The experimental results provide feedback on the realization bottlenecks and indicate that when the right configurable tools are selected, AVS encoder will be suitable for application of mobile video communications.
- Realization of the detection method for human falling action in room based on embedded system1
The experiment result shows that resolution of falling-action detection for human based on vision and embedded system has reached the feature of real-time and accuracy for the detection system.
- AAML Based Avatar Animation with Personalized Expression for Online Chatting System1
Successful application of the online chatting system shows that the system can enhance affective interaction by the customized facial animation and the animated avatar that is similar with the user.
- Learn Traffic State Based on Cooperative Localization1
This paper presents a new cooperative localization technique which makes use of collaboration of mobile phones for traffic monitoring and utilizes additional connection information among mobile phones to improve the localization accuracy.
- Fast rate-distortion algorithm based on quasi-zero-block detection in AVS-M1
The method adds computation for Square Sum of Distortion of non-zero blocks, but it can omit about 70%-90% computation for transform,quantization, inverse transform,verse transform, inverse quantization and reconstruction with little video quality degradation.
- A Kernel-Based Case Retrieval Algorithm with Application to Bioinformatics1
An approach to utilizing the correlative information among features to compute the similarity of cases for case retrieval is provided by extending the dot product-based linear similarity measures to their nonlinear versions with kernel functions.
- Intelligent Protein 3D Structure Retrieval System1
An intelligent protein 3D structure retrieval system that integrates the moment feature extraction technology and the relevant feedback method in Artificial Intelligence (AI) to aid the biologists to find the similar protein structures more effectively.
- FLASHand: Feed-forward reLightable and Animatable Single-view Hand Reconstruction–
FLASHand is proposed, the first feed-forward relightable and animatable 3D hand reconstruction model from a single RGB image, and introduces a mesh-based disentangled 2D Gaussian splatting hand representation to ensure plausible geometry generation and achieve high-fidelity animation and relighting.
- –
- –
- –
- VersaFusion: A Versatile Diffusion-Based Framework for Fine-Grained Image Editing and Enhancement–
Experimental results demonstrate the effectiveness of the proposed approach in enhancing the quality and controllability of T2I-generated images, surpassing existing methods while maintaining attractive computational complexity for practical real-world applications.
- FHBench: Towards Efficient and Personalized Federated Learning for Multimodal Healthcare–
The results highlight the robustness of FHBench as a benchmarking tool and the potential of EPFL as an innovative approach to advancing healthcare-focused FL, addressing key limitations of existing methods.
- –
- Predicting Intraocular Collamer Lens Vault in Myopic Patients With Shallow Anterior Chamber Using FedEYE Platform and ChatGPT–
These two code-free AI models demonstrated encouraging performance in predicting vault in shallow ACD patients as well as using different AI models to predict vault after ICL implantation could be of clinical significance for improving surgical safety and efficacy in shallow ACD patients.
- –
- –
- –
- –
- –
- –
- –
- Differentially Private Pre-Trained Model Fusion using Decentralized Federated Graph Matching–
This work proposes PrivFusion, a novel architecture that preserves privacy while facilitating model fusion under the constraints of local differential privacy, and incorporates a hybrid local differentially private mechanism and decentralized federated graph matching, effectively protecting both activation values and weights.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Tap2Pair–
Tap2Pair allows users to simply tap on an advertising device to associate with the target scanning device, and does not require any modification of existing wireless devices and is compatible with most wireless protocols.
- –
- Comprehensive Data Management and Analytics for General Society Survey Dataset–
This paper proposes a comprehensive data management and data mining approach for GSS datasets which can explore knowledges in a fine-grained manner with minimum human interference and leverages the power of data mining techniques.
- –
- –
- –
- –
- –
- Recognizing extended surrounding contexts via class incremental learning–
Results of the experiments show that FCI-ELM can automatically and continuously recognize newly discovered classes of contexts in the real-world, and the model can keep the same recognition ability by class incremental learning approach.
- Text to Head Motion Synthesis for Chinese Sign Language Avatar–
A quintuple for formalized head movement description is proposed by analyzing the head motions both defined in words of CSL and captured from large motion data of a real signer performance to perform a low-level semantic mapping from words to head gestures.
- –
- –
- Game Theoretic Analysis for Cooperative Video Transmission over Heterogeneous Devices–
A novel power-allocation model of the base station (BS) based on noncooperative game theory and bidding is presented in this study, and a utility function of Signal-to-Noise Ratios (SNRs) along with Signal- to-Interference Ratio (SIRs) is proposed.
- CaneFitter: investigation on appropriate cane selection and proper cane use for the elderly–
CaneFitter is presented, a cane fitting system that consists of two major components: 1) recommendation module, which generates adaptive recommendations on better fitting canes based on the efficient gait analysis technique; 2) teaching module, where the image of an elderly user segmented from the live video is merged with the training video, allowing the user to learn the cane use in an immersive augmented virtual environment.
- –
- –
- –
- SAP dissimilarity based high performance Wi-Fi indoor localization–
A new dissimilarity based localization method SAP (Signal-AP) is proposed to implement high performance indoor localization and the results show that SAP not only improves the localization accuracy, but also has desirable scalability of environment.
- –
- –
- –
- Dynamic collaborative indoor locating method based on multiple terminals–
Experimental results show that the proposed dynamic collaborative indoor locating method not only realizes high precision positioning, but also can avoid repetitive calibration, beneficial to actual application.
- 3D Gesture Recognition Scheme Based on Gyro Sensor–
Experimental results show that the average recognition rate of this scheme is 99.3% which can well realize 3D space mouse input function and the keyboard function.
- Building of pervasive multimedia message distribution system–
An efficient multimedia distribution system based on LOONGSON-2F is designed, and a novel schedule problem for distributing the content based on the request attributes is presented, which meets the user requirement.
- Study of building low-cost pervasive network and optimizing its services–
This pa-per proposes a user-property-based access algorithm and a priority-based bandwidth allocation algorithm that exhibit better performance than average-allocation algorithm, in term of efficiency and service time.
- –
- MMS entertainment system based on mobile phone–
This work proposed automatic caricature generation method based on Manifold Regularization and speech driven animation approach based on Hidden Markov Model, which can provide people attractive MMS for mobile entertainment successfully.
- Probability matching based on historical path in location estimation–
This paper proposes a shortest history path algorithm based on probability matching, called ProHI, which not only works in the singular AP environment, but also performs well when multiple APs are used, and compares with the traditional algorithm.
- Complex constellation graph of multivariate data and its optimization method–
The proposed visual analysis method, which allow the weights to be complex numbers and determine the weights automatically based on complex linear discriminant analysis of the data, indicates that the method reveal the structural relationship of original data well and can be used combining machine algorithms as a tool for visual data analysis.
- User Behavior Mining for On-Line GUI Adaptation–
A hierarchical Markov model is proposed for mining and predicting user's behavior from Wiimote control sequence and the modal can be on-line updated and highlight the next possible operation and then improve the system's usability.
- To Probe into the Information Resource Construction of University Library——From the Point of View of Economics–
The author expects to find available measures which can bring the information resource of university library into the important play in the aspect of economical benefit and society benefit of this country.
- C-E Translation of Abstracts of Metallurgy Science Papers–
Appropriate translation strategies should be adopted in its English rendition to make the target text enjoy natural expression, logical arrangement and precise diction on the basis of a thorough understanding of the pragmatic, logic and syntactic distinctions between Chinese and English science texts.
- Application of modified Hausdorff distance in face match–
A new method to match human faces with the Modified Hausdorff Distance is presented and it is shown that this method performs well while retrieving face from the database according to a user’s input face.
- –
- –
- SSC: Gesture-based game for initial dementia examination–
A novel system assisting medical dementia examination in a joyful way: the object just needs to play a popular game SSC against the computer during the examination and can improve people’s mental status.
- Implementation of AVS-M Decoder on BREW Platform–
The method for transplanting reference code of AVS-M decoder to BREW platform and problems that need to be solved are introduced and a few methods for optimizing reference code are given.
- –
- Open Issues on Intelligent Sensor Networks–
Some open issues on intelligent sensor networks research are addressed in this paper, which addresses some open issues in wireless communications and electronics research.
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.