A5013563341- Advanced Neural Network Applications
- Material Dynamics and Properties
- Nanopore and Nanochannel Transport Studies
- Electrostatics and Colloid Interactions
- Advanced Data Compression Techniques
- Neural Networks and Applications
- Advanced SAR Imaging Techniques
- Generative Adversarial Networks and Image Synthesis
- Fuel Cells and Related Materials
- Membrane-based Ion Separation Techniques
- Robotics and Sensor-Based Localization
- Adversarial Robustness in Machine Learning
- Spectroscopy and Quantum Chemical Studies
- Domain Adaptation and Few-Shot Learning
- Model Reduction and Neural Networks
China Academy of Space Technology
2021-2022
Academy of Mathematics and Systems Science
2016-2017
National Center for Mathematics and Interdisciplinary Sciences
2016-2017
University of Chinese Academy of Sciences
2017
Chinese Academy of Sciences
2016
Deep learning methods have made significant progress in ship detection synthetic aperture radar (SAR) images. The pretraining technique is usually adopted to support deep neural networks-based SAR detectors due the scarce labeled However, directly leveraging ImageNet hard obtain a good detector because of different imaging perspectives and geometry. In this paper, resolve problem inconsistent between earth observations, we propose an optical (OSD) transfer characteristics ships observations...
This paper proposes a Kolmogorov high order deep neural network (K-HOrderDNN) for solving high-dimensional partial differential equations (PDEs), which improves the networks (HOrderDNNs). HOrderDNNs have been demonstrated to outperform conventional DNNs frequency problems by introducing nonlinear transformation layer consisting of $(p+1)^d$ basis functions. However, number functions grows exponentially with dimension $d$, results in curse dimensionality (CoD). Inspired superposition theorem...
Potassium channels are much more permeable to potassium than sodium ions, although ions larger and both carry the same positive charge. This puzzle cannot be solved based on traditional Poisson-Nernst-Planck (PNP) theory of electrodiffusion because PNP model treats all as point charges, does not incorporate ion size information, therefore discriminate from ions. The can qualitatively capture some macroscopic properties certain channel systems such current-voltage characteristics, conductance...
The energy functional, the governing partial differential equation(s) (PDE), and boundary conditions need to be consistent with each other in a modeling system. In electrolyte solution study, people usually use free form of an infinite domain system (with vanishing potential condition) derived PDE(s) for analysis computing. However, many real systems and/or numerical computing, objective is bounded, still similar form, PDE(s), but different conditions, which may cause inconsistency. this...
Generative adversarial networks (GANs) have attracted in-tense interest in the field of generative models. This paper will first theoretically analyze GANs' approximation property. Similar to universal property fully connected neural with one hidden layer, we prove that generator input latent variable GANs can universally approximate potential data distribution given increasing neurons. Furthermore, propose an approach named stochastic generation (SDG) enhance ability. Our is based on simple...