A5101556659- Thermography and Photoacoustic Techniques
- Video Surveillance and Tracking Methods
- Reinforcement Learning in Robotics
- Visual Attention and Saliency Detection
- Infrared Target Detection Methodologies
- Advanced Neural Network Applications
- Natural Language Processing Techniques
- Advanced Image and Video Retrieval Techniques
- Advanced Measurement and Detection Methods
- Neural Networks and Applications
- Speech and dialogue systems
- Welding Techniques and Residual Stresses
- Topic Modeling
- Calibration and Measurement Techniques
- Neural dynamics and brain function
- Model Reduction and Neural Networks
- Corporate Social Responsibility Reporting
- Ultrasonics and Acoustic Wave Propagation
- Fire Detection and Safety Systems
- Artificial Intelligence in Games
- Gaze Tracking and Assistive Technology
- Fire effects on concrete materials
- Graph Theory and Algorithms
- Wireless Signal Modulation Classification
- Remote-Sensing Image Classification
Chinese University of Hong Kong, Shenzhen
2024
Chengdu University of Information Technology
2024
University of Washington
2024
Xidian University
2019-2022
The University of Texas at Dallas
2017
Air Force Engineering University
2004-2017
Beijing University of Chinese Medicine
2016
Xi'an High Tech University
2011-2013
Nanjing Second Hospital
2013
Horological Research Institute of Light Industry
2010
Most existing deep reinforcement learning (DRL) frameworks consider either discrete action space or continuous solely. Motivated by applications in computer games, we the scenario with discrete-continuous hybrid space. To handle space, previous works approximate discretization, relax it into a set. In this paper, propose parametrized Q-network (P- DQN) framework for without approximation relaxation. Our algorithm combines spirits of both DQN (dealing space) and DDPG seamlessly integrating...
Starcraft II (SC2) is widely considered as the most challenging Real Time Strategy (RTS) game. The underlying challenges include a large observation space, huge (continuous and infinite) action partial observations, simultaneous move for all players, long horizon delayed rewards local decisions. To push frontier of AI research, Deepmind Blizzard jointly developed StarCraft Learning Environment (SC2LE) testbench complex decision making systems. SC2LE provides few mini games such MoveToBeacon,...
Pedestrian detection is a crucial component for intelligent transport system and advanced driver assistance system. In recent years, pedestrian methods have achieved higher accuracy. However, the existing algorithms are insufficient small-scale that relatively far from cameras in practical applications. this paper, we propose novel deep sense network (termed SSN) detection. The proposed architecture could generate some proposal regions which more effective to detect pedestrians. Furthermore,...
The environment of sports competition is changing rapidly, and there a certain relationship between athletes’ decision-making executive functions emotions. Positive emotions can enhance reaction inhibition, while negative will damage the inhibition function. Therefore, identifying athletes in competitions help coaches quickly grasp emotional state athletes, so as to make targeted decisions. With advent era big data continuous in-depth development deep neural networks, emergence various...
Ships comprise the only and most important ocean transportation mode. Thus, ship detection is one of critical technologies in monitoring, which plays an essential role maintaining marine safety. Optical remote-sensing images contain rich color texture information, beneficial to detection. However, few optical datasets are open publicly due issue sensitive data copyrights, HRSC2016 dataset built for ship-detection task. Moreover, almost all general object detectors suffer from failure...
Voice-based human-machine interfaces are becoming a key feature for next generation intelligent vehicles. For the navigation dialogue systems, it is desired to understand driver's spoken language in natural way. This study proposes two-stage framework, which first converts audio streams into text sentences through Automatic Speech Recognition (ASR), followed by Natural Language Processing (NLP) retrieve navigation-associated information. The NLP stage based on Deep Neural Network (DNN)...
Abstract Recurrent neural networks (RNN) are ubiquitous computing systems for sequences and multivariate time-series data. While several robust RNN architectures known, it is unclear how to relate initialization, architecture, other hyperparameters with accuracy a given task. In this work, we propose treating as dynamical correlating through Lyapunov spectral analysis, methodology designed explicitly nonlinear systems. To address the fact that features go beyond existing infer relevant from...
Voice-based human-machine interface has become a prevalent feature for modern intelligent vehicles, especially in navigation and infotainment applications. Automatic Speech Recognition (ASR) converts spoken audio streams to plain texts, but follow-up Natural Language Processing (NLP) sub-system is needed understand the contextual meaning from text act. For specific human-vehicle dialogue application, two major tasks include (1) intent detection - decide whether sentence navigation-related,...
Frequency hopping (FH) signal is widely adopted by military communications as a kind of low probability interception signal. Therefore, it very important to research the FH detection algorithm. The existing algorithm signals based on time-frequency analysis cannot satisfy time and frequency resolution requirement at same due influence window function. In order solve this problem, an wavelet decomposition Hilbert–Huang transform (HHT) was proposed. proposed removes noise received detects...
Pulse diagnosis is an important part of the theoretical system Traditional Chinese Medicine. In this paper we propose a new and efficient framework to recognize pulse signal in nonlinear angle. Firstly EEMD (ensemble empirical mode decomposition) method used filter out baseline drifting noise, result proved be effective. Then MFDFA(multi-fractal detrended fluctuation analysis) get Hurst index, Renyi index multi-fractal spectrum. related with long-range correlations, characteristics, spectrum...
Random Recurrent Neural Networks (RRNN) are the simplest recurrent networks to model and extract features from sequential data. The simplicity however comes with a price; RRNN known be susceptible diminishing/exploding gradient problem when trained gradient-descent based optimization. To enhance robustness of RRNN, alternative training approaches have been proposed. Specifically, FORCE learning approach proposed recursive least squares train was shown applicable even for challenging task...
Retracted.
Aims at the safety problems caused by corrosion under coating for it can not easily be observed, specimen of damage was made research detection applying infrared thermal wave technology. The surface temperature distribution images were captured, also history area, non-corrosion area and their difference. Analysis test results show that technology is an effective tool in detecting coating, there has a fine time used to estimate thickness, position sizes defects found out directly hot spots...
Defects in the specimen of missile engine shell was detected by thermal wave image technology this paper. In order to gain intuitionistic and accurate space structure object, subtracting background high-frequency emphasized filtering method were used enhance quality. Then defect segmented from using particle swarm fuzzy clustering algorithm, while size depth identified quantitatively. On basis, 3D reconstruction recognized Volume Rendering method. The results show that precision quantitative...
For the needs of fast and efficient detection crack defects ceramic material, ultrasonic infrared thermal wave method was applied for research. Ultrasonic excitation is on a bowl which containing defect, image obtained by camera. The experimental results show that suitable damage materials, speed with time less than 1s visual results.