A5012433446- Reinforcement Learning in Robotics
- Logic, Reasoning, and Knowledge
- AI-based Problem Solving and Planning
- Semantic Web and Ontologies
- Domain Adaptation and Few-Shot Learning
- Cloud Computing and Resource Management
- Adaptive Dynamic Programming Control
- Smart Grid Energy Management
- Robotic Path Planning Algorithms
- Natural Language Processing Techniques
- Topic Modeling
- Bayesian Modeling and Causal Inference
- Machine Learning and ELM
- Multi-Agent Systems and Negotiation
- Advanced Neural Network Applications
- Generative Adversarial Networks and Image Synthesis
- Robotics and Sensor-Based Localization
- Evolutionary Algorithms and Applications
- Numerical methods in inverse problems
- Logic, programming, and type systems
- Distributed and Parallel Computing Systems
- Multimodal Machine Learning Applications
- Graph Theory and Algorithms
- Cloud Data Security Solutions
- Data Management and Algorithms
Jilin University
2016-2025
Jiangyin People's Hospital
2025
University Town of Shenzhen
2024
Tsinghua University
2024
Nanchang Hangkong University
2023-2024
Jilin Medical University
2013-2024
Henan University of Science and Technology
2024
Zhejiang Sci-Tech University
2024
Microsoft Research Asia (China)
2023
Jilin Province Science and Technology Department
2006-2022
Recently, instance contrastive learning achieves good results in unsupervised domain adaptation. It reduces the distances between positive samples and anchor, increases negative learns discriminative feature representations for target samples. However, most recent methods identifying are based on whether pseudo-labels of pseudo-label anchor correspond to same class. Due lack labels, many uncertain data mistakenly labeled during training process, low potential also utilized. To address these...
For addressing the data privacy and portability issues of domain adaptation, Domain Adaptation Black-box Predictors (DABP) aims to adapt a black-box source model an unlabeled target without accessing both source-domain details model. Although existing DABP approaches based on knowledge distillation (KD) have achieved promising results, we experimentally find that these methods all minority class forgetting issue, which refers trained completely forgets some classes. To address this propose...
Generative Adversarial Networks (GAN) is an adversarial model, and it has been demonstrated to be effective for various generative tasks. However, GAN its variants also suffer from many training problems, such as mode collapse gradient vanish. In this paper, we firstly propose a general crossover operator, which can widely applied GANs using evolutionary strategies. Then design framework named C-GAN based on it. And combine the operator with networks (E-GAN) implement (CE-GAN). Under premise...
The recommender system can help users solve the problem of information overload and find item which user requires efficiently. In this paper, we combine text image in a given user's browsing news article classify through deep neural network, then recommend article's tag to user. process extracting eigenvector, apply Convolutional Neural Network(CNN) method because its good performance. On other hand, VGG is used extract eigenvector. Meanwhile imply Autoencoder(AE) reduce dimension output...
Ionospheric tomography based on the observed total electron content along different satellite-receiver rays is a typically ill-posed inverse problem. Incorporating density profiles data from COSMIC radio occultation technique and ground ionosondes, Tikhonov regularization method adopted to reconstruct 3-D ionospheric density, parameter used balance weights between prior (or background) information real measurements. To determine optimal parameter, model function in modified L-curve used....