A5001711262- Advanced Fiber Laser Technologies
- IoT and Edge/Fog Computing
- Context-Aware Activity Recognition Systems
- Human Pose and Action Recognition
- Laser-Matter Interactions and Applications
- Advanced Computing and Algorithms
- Solid State Laser Technologies
- Photonic Crystal and Fiber Optics
- Emotion and Mood Recognition
- Recommender Systems and Techniques
- Optical Imaging and Spectroscopy Techniques
- Advanced Fiber Optic Sensors
- Caching and Content Delivery
- Image and Object Detection Techniques
- Green IT and Sustainability
- Hand Gesture Recognition Systems
- Semiconductor Lasers and Optical Devices
- Gaze Tracking and Assistive Technology
- Industrial Vision Systems and Defect Detection
- Psychosocial Factors Impacting Youth
- Photonic and Optical Devices
- Anomaly Detection Techniques and Applications
- Non-Invasive Vital Sign Monitoring
- Surface Roughness and Optical Measurements
Beijing Electronic Science and Technology Institute
2022-2024
University of Chinese Academy of Sciences
2024
Nanjing Normal University
2021
Institute of Psychology, Chinese Academy of Sciences
2021
Czech Academy of Sciences, Institute of Psychology
2020
Chinese Academy of Sciences
2020
Collaborative Innovation Center of Advanced Microstructures
2017-2018
Nanjing University
2017-2018
Due to rapid development of sensor technology, human activity recognition (HAR) using wearable inertial sensors has recently become a new research hotspot. Deep learning, especially convolutional neural network (CNN) that can automatically learn intricate features have gained lot attention in ubiquitous HAR task. Most existing CNNs process input by extracting channel-wise features, and the information from each channel be separately propagated hierarchical way lower layers higher layers. As...
Recently, the state-of-the-art performance in various sensor-based human activity recognition (HAR) tasks has been acquired by deep learning, which can extract automatically features from raw data. In standard convolutional neural networks (CNNs), there is usually same receptive field (RF) size of artificial neurons within each feature layer. It well known that RF able to change adaptively according stimulus, rarely exploited HAR. this article, a new multibranch CNN introduced, utilizes...
WTe2 is a unique material in the family of transition metal dichalcogenides and it has been proposed as candidate for type-II Weyl semimetals. However, thus far, studies on optical properties this emerging have significantly hindered by lack large-area, high-quality materials. Here, we grow centimeter-scale, highly crystalline ultrathin film (∼35 nm) pulsed laser deposition technique. Broadband pump-probe spectroscopy (1.2–2.5 μm) reveals peculiar ultrafast response where an initial...
Although automatic emotion recognition from facial expressions and speech has made remarkable progress, body gestures not been thoroughly explored. People often use a variety of language to express emotions, it is difficult enumerate all emotional collect enough samples for each category. Therefore, recognizing new critical better understanding human emotions. However, the existing methods fail accurately determine which state gesture belongs to. In order solve this problem, we introduce...
We have for the first time experimentally demonstrated a high repetition-rate picosecond fiber laser source at 2 μm by spectrally masked phase modulation technique, where modulator driven sinusoidal RF signal and Bragg grating are used to convert output of 2-μm continuous-wave single-longitudinal-mode diode pulse train. The this can be continuously flexibly tuned from 1 6 GHz simply changing signal. achieved shortest width ~60 ps SNR >75 dB an operating frequency GHz. simplicity robustness...
Human activity recognition (HAR) in ubiquitous computing has been beginning to incorporate attention into the context of deep neural networks (DNNs), which rich sensing data from multimodal sensors such as accelerometer and gyroscope is used infer human activities. Recently, two methods are proposed via combining with Gated Recurrent Units (GRU) Long Short-Term Memory (LSTM) network, can capture dependencies signals both spatial temporal domains simultaneously. However, recurrent often have...
Objective: Explore the relationship between parental support, hope, and learning engagement in secondary school students, as well mediating mechanisms of support on engagement. Method: This study selected students from vocational schools participants conduct
In this paper, we introduce MP-GT, a novel Graph Neural Network model that leverages meta-path-guided optimization within the GCN-Transformer framework to enhance application (App) usage prediction. Our approach addresses issues such as suspended animation and over-smoothing by extracting both local subgraph structures global graph using GT method. Furthermore, capture of semantic information App patterns incorporating meta path-guided objective function. Extensive experiments demonstrate...
We report that MBE-grown three-dimensional (3D) topological Dirac semimetal Cd3As2 thin-film exhibits remarkable saturable absorption effects at 1, 1.5 and 2 μm. A mode-locked Tm fiber laser is demonstrated using such a SESAM-like material.
Eye movement (EM), as a mature observation technology, has been widely used in the research of psychology, and it is also one important methods multi-quality psychological testing technology. However, there are relatively few researches on based EM technology at present. By introducing convolution neural network (CNN) into deep long short memory (DLSTM), this paper develops new structure, designs fusion strategy, proposes an tracking data algorithm learning (EYE-CNN-DLSTM). comparing effect...
As convolutional neural network (CNN) has the parallel pooling layer and a large number of multiplicative operations, it becomes hot research topic to accelerate forward operation by using field programmable gate array (FPGA). This paper proposes design implementation method an FPGA-based CNN accelerator. Based on Xilinx ZYNQ, this selects typical model, ARM+FPGA software-hardware collaboration scheme which is controlled ARM accelerated FPGA hardware. Through collaboration, adopts...
Spectrally masked phase modulation technique is used to demonstrate a 2 μm picosecond source, with continuously tunable repetition rate up 6 GHz. Such source useful for data communication and processing.
We demonstrate a 15-GHz actively-mode-locked thulium fiber laser. The repetition-rate is improved by one order of magnitude compared with existing results and such source can be used for 2 µm optical data-communication processing.
Surface defect inspection based on machine vision is often affected by uneven illumination. In order to improve the rate of surface defects under illumination condition, this paper proposes a method for detecting image convolutional neural network, which adjustment networks, training parameters, changing structure achieve purpose accurately identifying various defects. Experimental copper strip and steel images shows that network can automatically learn features without preprocessing image,...