A5101450514- Human Mobility and Location-Based Analysis
- Data Management and Algorithms
- CCD and CMOS Imaging Sensors
- Neural Networks and Applications
- Green IT and Sustainability
- Stochastic Gradient Optimization Techniques
- Parallel Computing and Optimization Techniques
- IoT and Edge/Fog Computing
- Time Series Analysis and Forecasting
- Advanced Neural Network Applications
- Advanced Steganography and Watermarking Techniques
- Advanced Memory and Neural Computing
- Handwritten Text Recognition Techniques
- Video Analysis and Summarization
- Geographic Information Systems Studies
- Context-Aware Activity Recognition Systems
- Diverse Aspects of Tourism Research
Xuzhou University of Technology
2025
Xidian University
2025
University of North Carolina at Chapel Hill
2019-2023
State Key Laboratory of Information Engineering in Surveying Mapping and Remote Sensing
2022-2023
Wuhan University
2022-2023
Tsinghua University
2017
We propose a novel text steganography method using RNN Encoder-Decoder structure to generate quatrains, one genre of Chinese poetry. Compared other text-generation based methods which have either very low embedding rate or flaws in the naturalness generated texts, our has higher and better quality. In this paper, we use LSTM model first line quatrain with keyword then following lines by one. proved effective generating poetry, but when applied steganograpy, poetry quality decreases sharply,...
Clustering the trajectories of vehicles moving on road networks is a key data mining technique for understanding human mobility patterns, as well their interactions with urban environments. The development efficient and scalable trajectory clustering algorithms, however, still faces challenges because computational costs when measuring similarities among large number network-constrained trajectories. To address this problem, novel framework based well-developed Density-Based Spatial...
The origin–destination (OD) matrix describes traffic flow information between regions. It is a critical input for intelligent transportation systems (ITS). However, obtaining the OD remains challenging due to high costs and privacy concerns. Synthetic data, which have same statistical distribution of real help address issues data scarcity. Based on Generative Adversarial Networks (GAN), generation models, can effectively generate synthetic matrix, challenge in ITS research. existing methods...
We present Antler, which exploits the affinity between all pairs of tasks in a multitask inference system to construct compact graph representation task set and finds an optimal order execution such that end-to-end time energy cost is reduced while accuracy remains similar state-of-the-art. The design Antler based on two observations: first, running same platform shows affinity, leveraged find helps avoid unnecessary computations overlapping subtasks set; second, run may have dependencies,...
Resource-optimized deep neural networks (DNNs) nowadays run on microcontrollers to perform a wide variety of audio, image and sensor data classification tasks. Despite comprehensive support for learning tools 32-bit microcontrollers, performing inferences 16-bit still remains chal-lenge. Although there are some implementing net-works systems, generally, is large gap in efficiency between the development (or higher) systems. There also steep curve that discourages beginners inexperienced with...