A5055922843- Face and Expression Recognition
- Adversarial Robustness in Machine Learning
- Topic Modeling
- Generative Adversarial Networks and Image Synthesis
- Advanced Image and Video Retrieval Techniques
- Face recognition and analysis
- Sentiment Analysis and Opinion Mining
- Robotic Path Planning Algorithms
- Domain Adaptation and Few-Shot Learning
- Image Retrieval and Classification Techniques
- Advanced Image Fusion Techniques
- Advanced Text Analysis Techniques
- Anomaly Detection Techniques and Applications
- Advanced Computing and Algorithms
- Multimodal Machine Learning Applications
- Human Pose and Action Recognition
- Video Surveillance and Tracking Methods
- Bacillus and Francisella bacterial research
- Digital Media Forensic Detection
- Advanced Image Processing Techniques
- Advanced Vision and Imaging
- Biometric Identification and Security
- Remote-Sensing Image Classification
- Image and Signal Denoising Methods
- Image and Video Quality Assessment
China University of Mining and Technology
2024
Southwest Jiaotong University
2024
Jiangnan University
2016-2024
Nanning Normal University
2020-2024
State Grid Corporation of China (China)
2024
Anhui Normal University
2024
National University of Defense Technology
2019-2023
Shanghai Academy of Agricultural Sciences
2021-2022
Chongqing University
2020-2021
Zhejiang University of Science and Technology
2021
Crop and aquatic animal co-culture systems have been used for over 1200 years can very efficiently utilize nutrients materials available in agroecology settings. These are sustainable forms of agriculture, extensively practiced worldwide. Microorganisms play critical roles promoting ecosystem nutrient transformations material circulations. However, the long-term effects this farming system on soil health microbial community stability, addition to their relationships with rice yields, remain...
By characterizing each image set as a nonsingular covariance matrix on the symmetric positive definite (SPD) manifold, approaches of visual content classification with sets have made impressive progress. However, key challenge unhelpfully large intraclass variability and interclass similarity representations remains open to date. Although, several recent studies mitigated two problems by jointly learning embedding mapping metric original SPD their inherent shallow linear feature...
In this paper, we propose a novel Deep Reinforcement Learning (DRL) algorithm which can navigate non-holonomic robots with continuous control in an unknown dynamic environment moving obstacles. We call the approach MK-A3C (Memory and Knowledge-based Asynchronous Advantage Actor-Critic) for short. As its first component, builds GRU-based memory neural network to enhance robot's capability temporal reasoning. Robots without it tend suffer from lack of rationality face incomplete noisy...
The construction of transportation 5.0 or the so-called society-centered intelligent systems (ITS) has aroused higher requirements for sensing capability to seamlessly integrate Cyber-Physical-Social Systems (CPSS). Crowd Sensing Intelligence (CSI), as a promising paradigm, leverages collective intelligence heterogeneous resources gather data and information from CPSS. Our first Distributed/Decentralized Hybrid Workshop on (DHW-CSI) been focused principles high-level processes organizing...
Natural soil and vegetation recovery following human disturbance is the primary means of restoring degraded ecosystems globally. However, it remains unclear how in tropical karst areas China affects physicochemical properties. Here, we investigated impacts natural on properties at different depths southwestern China, using a space–time substitution method. We found that with recovery, bulk density (SBD) decreased. Soil pH initially decreased then increased, reaching its lowest value during...
Environmental DNA (eDNA) metabarcoding has been widely used in freshwater systems, contributing to the advancements monitoring of fish diversity and community species composition. Nevertheless, accuracy reliability eDNA assessing functional structures revealing mechanisms underlying assembly remain unclear. In this study, we combined a traditional survey method (electrofishing) conduct stock upper reaches Huishui stream. We assessed taxonomic structures, as well mechanisms, during dry wet...
The improved generative adversarial network (improved GAN) is a successful method using model to solve the problem of semi-supervised learning (SSL). GAN learns generator with technique mean feature matching which penalizes discrepancy first-order moment latent features. To better describe common attributes distribution, this paper proposes novel SSL incorporates and secondorder moments features in an intermediate layer discriminator, called variance (MVFM-GAN). capture more precisely data...
To learn disentangled representations of facial images, we present a Dual Encoder-Decoder based Generative Adversarial Network (DED-GAN). In the proposed method, both generator and discriminator are designed with deep encoder-decoder architectures as their backbones. be more specific, structured is used to pose face representation, tasked perform real/fake classification, reconstruction, determining identity estimating pose. We further improve network architecture by minimizing additional...
Since an individual approach can hardly navigate robots through complex environments, we present a novel two-level hierarchical framework called JPS-IA3C (Jump Point Search improved Asynchronous Advantage Actor-Critic) in this paper for robot navigation dynamic environments continuous controlling signals. Its global planner JPS+ (P) is variant of JPS Search), which efficiently computes abstract path neighboring jump points. These nodes, are seen as subgoals, completely rid Deep Reinforcement...
Adversarial examples have raised great concerns about the security of deep learning models. Substitute training makes it possible to conduct black-box substitute attacks in real-world scenarios where attacker does not need access structure, parameters, and set target model. However, existing methods require a large number queries on model suffer from low attack success rates. To alleviate these problems, we propose novel adversarial method, named meta-learning (SML), which combines with...
Thanks to the efficacy of Symmetric Positive Definite (SPD) manifold in characterizing video sequences (image sets), image set-based visual classification has made remarkable progress. However, issue large intra-class diversity and inter-class similarity is still an open challenge for research community. Although several recent studies have alleviated above by constructing Riemannian neural networks SPD matrix nonlinear processing, degradation structural information during multi-stage...
Fault diagnosis and particle discrimination can be fundamentally solved as a case of pulse shape (PSD). The classical methods PSD are inconvenient or not effective when more than two shapes need to discriminated the have only small differences. A direct method discriminate nuclear based on principal component analysis (PCA) support vector machine (SVM) is reported in this paper. training testing accuracies SVM classifiers with different kernels were same, algorithms shown great noise...