A5100357386- Higher Education and Teaching Methods
- Advanced Computational Techniques and Applications
- Music and Audio Processing
- Generative Adversarial Networks and Image Synthesis
- Data Mining Algorithms and Applications
- Semantic Web and Ontologies
- Advanced Image and Video Retrieval Techniques
- Advanced Neural Network Applications
- Multimodal Machine Learning Applications
- Image Retrieval and Classification Techniques
- Handwritten Text Recognition Techniques
- Rough Sets and Fuzzy Logic
- Speech Recognition and Synthesis
- Library Collection Development and Digital Resources
- Advanced Memory and Neural Computing
- Emotion and Mood Recognition
- Speech and Audio Processing
- Data Management and Algorithms
- Domain Adaptation and Few-Shot Learning
- EEG and Brain-Computer Interfaces
- Advanced Algorithms and Applications
- Neural dynamics and brain function
- Natural Language Processing Techniques
- Technology and Security Systems
- Underwater Vehicles and Communication Systems
Yanshan University
2014-2024
Guangdong University of Technology
2024
Zhejiang University
2004-2022
Henan University of Technology
2022
National Engineering Research Center for Wheat
2022
South China University of Technology
2021
Tencent (China)
2018-2019
China Electric Power Research Institute
2017
National University of Defense Technology
2013
Second Military Medical University
2010
In existing visual representation learning tasks, deep convolutional neural networks (CNNs) are often trained on images annotated with single tag, such as ImageNet. However, tag annotation cannot describe all important contents of one image, and some useful information may be wasted during training. this work, we propose to train CNNs from multiple tags, enhance the quality CNN model. To end, build a large-scale multi-label image database 18M 11K categories, dubbed <italic...
In this work we propose a new automatic image annotation model, dubbed diverse and distinct (D2IA). The generative model D2IA is inspired by the ensemble of human annotations, which create semantically relevant, yet tags. D2IA, generate relevant tag subset, in tags are to contents each other, using sequential sampling from determinantal point process (DPP) model. Multiple such subsets that cover semantic aspects or levels generated randomly perturbing DPP process. We leverage adversarial...
Decidualization refers to the process of transformation endometrial stromal fibroblast cells into specialized decidual that provide a nutritive and immunoprivileged matrix essential for blastocyst implantation placental development. Deficiencies in decidualization are associated with variety pregnancy disorders, including female infertility, recurrent failure (RIF), miscarriages. Despite increasing number genes reportedly receptivity decidualization, cellular molecular mechanisms triggering...
Speech emotion recognition is a vital contributor to the next generation of human-computer interaction (HCI). However, current existing small-scale databases have limited development related research. In this paper, we present LSSED, challenging large-scale english speech dataset, which has data collected from 820 subjects simulate real- world distribution. addition, release some pre-trained models based on can not only promote recognition, but also be transferred downstream tasks such as...
To resolve the difficulty in establishing accurate priori noise model for extended Kalman filtering algorithm, propose fractional-order Darwinian particle swarm optimization (PSO) algorithm has been proposed and introduced into fuzzy adaptive algorithm. The natural selection method adopted to improve standard which enhanced diversity of particles avoided premature. In addition, fractional calculus used evolution speed particles. PSO after improved applied train filter achieve simultaneous...
Spike sorting is a fundamental step in extracting single-unit activity from neural ensemble recordings, which play an important role basic neuroscience and neurotechnologies. A few algorithms have been applied spike sorting. However, when noise level or waveform similarity becomes relatively high, their robustness still faces big challenge. In this study, we propose method combining Linear Discriminant Analysis (LDA) Density Peaks (DP) for feature extraction clustering. Relying on the joint...
The output power of a photovoltaic system largely depends on the amount solar radiation that can be received by panel, and energy reaching ground is affected transmission process. However, in engineering practice, numerical simulation prediction schemes tend to adopt kind scheme, cannot always meet accuracy. In this paper, NCEP–NCAR reanalysis data are used as initial field, variety parameterization produce simulations for Xinjiang area. Through analysis examples, it found results differ...
Traditional machine learning-based methods for the detection of rice degree milling (DOM) that are not comprehensive in feature extraction and have low recognition rates fail to meet demand fast, non-destructive, accurate detection. This paper presents a digital image processing technology combined with deep learning implement classification DOM rice. An improved multi-scale information fusion model InceptionResNet–Bayesian optimization algorithm (IRBOA) was constructed based on Inception-v3...
<title>Abstract</title> Addressing the challenges of complex backgrounds, small parabolic targets, indistinct appearance features parabolas, and easy loss tracking in high-altitude object detection, this paper proposes a hybrid approach integrating Mixture Gaussians Background Modeling (MOG) algorithm with neural networks an improved Simple Online Realtime Tracking (SORT) for tracking. Firstly, to mitigate issues target parabolas region-specific conditional filtering is introduced reduce...
Data mining and knowledge discovery in data streams have recently attracted more attentions for their applications to numerous types of data, including Web clickstreams, sensor networks, etc. Because some special characteristics, such as continuous arrival multiple, rapid, time-varying, possibly unpredictable unbounded, yielded fundamentally new research problems. Among the various topics this field, it is paramount find frequent patterns a single pass, or small number passes, while using...
In this work we propose a new automatic image annotation model, dubbed {\bf diverse and distinct annotation} (D2IA). The generative model D2IA is inspired by the ensemble of human annotations, which create semantically relevant, yet tags. D2IA, generate relevant tag subset, in tags are to contents each other, using sequential sampling from determinantal point process (DPP) model. Multiple such subsets that cover semantic aspects or levels generated randomly perturbing DPP process. We...
"Semantic gap" is the major bottleneck of semantic-based multimedia retrieval technique in field information retrieval. Studies have shown that robust image can be achieved by single-trial visual evoked event related potential (ERP) detection. However, question remains whether auditory ERP utilized to achieve sound In this paper, we investigated rapid serial presentation (RSAP) paradigm. Eight BCI-naïve participants were instructed perform target detection RSAP sequences with vocalizations 8...
This paper presents a model of literature resource portal based on virtual and dynamic hierarchical architecture (VDHA) in an e-Science Grid-CUEG. Literature resources are important to the scientists. How exactly fully access distributed heterogeneous is big issue. avoid disadvantage centralizing index server's crashing down another The VDHA we proposed here can solve these problems.
In this paper we consider aggregate predicates and their support in database systems. Aggregate are the predicate equivalent to functions that they can be used search for tuples satisfy some property over a set of (as opposed simply computing an tuples). The importance is exemplified by many modern applications require ranked search, or top-k queries. Such queries norm multimedia spatial databases.In order concept DBMS, introduce several extensions query language engine. Specifically, extend...