A5100566996- Advanced Neural Network Applications
- Complex Network Analysis Techniques
- Opinion Dynamics and Social Influence
- Integrated Circuits and Semiconductor Failure Analysis
- Advanced Sensor and Control Systems
- Fire Detection and Safety Systems
- Video Surveillance and Tracking Methods
- Robotics and Sensor-Based Localization
- Physical Unclonable Functions (PUFs) and Hardware Security
- Image and Object Detection Techniques
- Advanced Measurement and Detection Methods
- Time Series Analysis and Forecasting
- Air Quality and Health Impacts
- Electrostatic Discharge in Electronics
- Noise Effects and Management
- Transportation Safety and Impact Analysis
- Stock Market Forecasting Methods
- Anomaly Detection Techniques and Applications
- Handwritten Text Recognition Techniques
- Electromagnetic Launch and Propulsion Technology
- Urban Transport and Accessibility
- Advanced Vision and Imaging
- Advanced Text Analysis Techniques
- Advanced Optical Sensing Technologies
- Opportunistic and Delay-Tolerant Networks
Nanjing University of Science and Technology
2023-2025
Communication University of China
2024
University of Eastern Finland
2024
Guizhou University
2022
Northwest A&F University
2022
Hunan Normal University
2020
Accurate prediction of financial market can promote the steady development market, but high frequency and noise time series make accurate a challenging task. In this paper, bidirectional long short-term memory neural network (BiLSTM) in deep learning is applied series, BiLSTM has one more layer reverse structure to try mine effective information. The performances unidirectional (LSTM), support vector regression (SVR) differential autoregressive moving average model (ARIMA) are compared....
Evidence on the link between long-term ambient particulate matter (PM) exposures and childhood sleep disorders were scarce. We examined associations to PM
The proliferation of multi-platform network information has expanded communication channels for users, enabling the integration and dissemination across both Social Networking Services (SNS)-type app Instant Message (IM)-type app. With intensification convergent communication, some users in two types apps show active alternation spreading to each other's platforms. study evolution trend different platforms is great practical significance mastery law. This synthesizes following three points:...
Aiming at the problem of false alarm in PCB defect detection automatic optical inspection process, many researchers have proposed their methods, but most them only classify single image, and there are multiple defects categories image. In this paper, a real data set consisting 1540 images generated by AOI is introduced for classification task. addition, we propose an improved detector based on feature pyramid networks. The combines Faster R-CNN FPN as infrastructure, has been adjusted...
Deep-learning techniques have significantly improved object detection performance, especially with binocular images in 3D scenarios. To supervise the depth information stereo detection, reconstructing dense of LiDAR point clouds causes higher computational costs and lower inference speed. After exploring intrinsic relationship between implicit semantic texture features images, we propose an efficient accurate algorithm, FCNet, images. First, construct a multi-scale cost-volume containing...
In the current firearms dynamics research, majority of studies focus on extraction resistance and subsequent recoil forces, with relatively fewer unlocking force. This study focuses measurement firearm force by analyzing previously designed device. By employing theoretical dynamic simulation analyses, variations over time were determined. A comparison two sets curves showed a high degree coincidence, indicating feasibility device in principle. Subsequently, to verify accuracy measurements...
In order to study the impact of matching muzzle brake and buffer device on disturbance a special firearm, this paper takes firearm as research object, based method rigid-flexible coupling dynamics, with platform for fixing flexible body. ADAMS dynamic software is used conduct simulation. The plan includes presence or absence brakes, buffering devices, devices different parameters, coordination between parameters brakes. simulations analyses aforementioned working conditions were proceeded...
Due to the high crowd density in some scenes and large differences feature scales, it is difficult count number of people. The structure multi-branch CNN can deal with features different improve accuracy counting, but also increase parameters become more complicated. In this paper, a 2S-ACNN simple balanced performance proposed for counting single frame image. network uses sizes receptive field adds multi-scale pooling carry out fusion, which representation ability model. experimental...