A5100426167- Advanced Graph Neural Networks
- Computational Drug Discovery Methods
- Machine Learning in Materials Science
- Neural Networks Stability and Synchronization
- Anomaly Detection Techniques and Applications
- Explainable Artificial Intelligence (XAI)
- Rough Sets and Fuzzy Logic
- Topic Modeling
- Nonlinear Dynamics and Pattern Formation
- Adversarial Robustness in Machine Learning
- Machine Learning and Data Classification
- Distributed Control Multi-Agent Systems
- Complex Network Analysis Techniques
- Recommender Systems and Techniques
- Domain Adaptation and Few-Shot Learning
- Imbalanced Data Classification Techniques
- Network Traffic and Congestion Control
- stochastic dynamics and bifurcation
- Face and Expression Recognition
- Advanced Image Fusion Techniques
- Generative Adversarial Networks and Image Synthesis
- Human Pose and Action Recognition
- Robotic Path Planning Algorithms
- Metaheuristic Optimization Algorithms Research
- Text and Document Classification Technologies
Chongqing University of Posts and Telecommunications
2014-2025
InferVision (China)
2025
Shanghai Electric (China)
2024
National University of Defense Technology
2023
State Key Laboratory of Modern Optical Instruments
2020
Zhejiang University
2020
Louisiana State University
2019
Qingdao National Laboratory for Marine Science and Technology
2019
Xiamen University
2019
China Earthquake Administration
2016-2018
Percutaneous transthoracic puncture of small pulmonary nodules is technically challenging. We developed a novel electromagnetic navigation system for the sub-centimeter lung by combining multiple deep learning models with and spatial localization technologies. compared performance DL-EMNS conventional CT-guided methods in percutaneous punctures using phantom animal models. In study, group showed higher technical success rate (95.6% vs. 77.8%, p = 0.027), smaller error (1.47 ± 1.62 mm 3.98...
Visual Instruction Tuning (VIT) enhances Multimodal Large Language Models (MLLMs) but it is hindered by corrupted datasets containing hallucinated content, incorrect responses, and poor OCR quality. While prior works focus on dataset refinement through high-quality data collection or rule-based filtering, they are costly limited to specific types of corruption. To deeply understand how affects MLLMs, in this paper, we systematically investigate issue find that while degrades the performance...
Aligning users across networks using graph representation learning has been found effective where the alignment is accomplished in a low-dimensional embedding space. Yet, highly precise remains challenging, especially for nodes with long-range connectivity to labeled anchors. To alleviate this limitation, we propose WL-Align which employs regularized framework learn distinctive node representations. It extends Weisfeiler-Lehman Isormorphism Test and learns alternating phases of...
A typical objective of bifurcation control is to delay the onset undesirable bifurcation. In this paper, problem Hopf in a second-order congestion model considered. particular, suitable created at desired location with preferred properties and dynamic delayed feedback controller developed for creation With controller, one can increase critical value communication delay, thus guarantee stationary data sending rate larger delay. Furthermore, explicit formulae determine period direction...
The deep convolutional neural networks (DCNN) require large number of training data to avoid overfitting, which makes it unsuitable for processing small-scale image datasets. transfer learning using DCNN (TCNN) reuses pre-trained layers generate a mid-level representation so that the optimization more than millions CNN parameters can be avoided. By this way, overfitting problem in alleviated. However, although now many public DCNNs have been trained and reused, existing TCNNs are formed by...
The smartphone has become an indispensable part in people's life. Identifying the user's emotional state according to usage of is a new way improve human-computer interaction and user experience. In this paper, we present attempt recognize states by using finger-stroke pattern. Firstly, International Affective Picture System (IAPS) were used design emotion inducing experiment. Then features under different categories extracted analyzed. Ultimately, machine learning algorithms identify three...
Pedestrian Attribute Recognition (PAR) has attracted increasing attention since it could provide important structural information of pedestrians for Smart Video Analysis. However, the pedestrian images are taken from a far distance significantly increase difficulty PAR fine-grained attributes. To address these problems, and further improve effects PAR, we proposed Multi-Scale Spatial Calibration (MSSC) module. More specifically, module includes two submodules: first, Calibrated Module (SCM)...