A5084988726- Traffic Prediction and Management Techniques
- Power Systems and Technologies
- Human Mobility and Location-Based Analysis
- IoT and Edge/Fog Computing
- Advanced Bandit Algorithms Research
- Distributed Sensor Networks and Detection Algorithms
- Automated Road and Building Extraction
- Data Visualization and Analytics
- Online Learning and Analytics
- Smart Grid and Power Systems
Shandong University
2024
Nanjing Institute of Railway Technology
2017-2023
Jiangsu Frontier Electric Technology Co., Ltd. (China)
2017-2023
Benefiting from instrumental global dependency modeling of self-attention (SA), transformer-based approaches have become the pivotal choices for numerous downstream visual reasoning tasks, such as question answering (VQA) and referring expression comprehension (REC). However, some studies recently suggested that SA tends to suffer rank collapse thereby inevitably leads representation degradation transformer layer goes deeper. Inspired by social network theory, we attempt make an analogy...
There are many problems caused by more and electric power big data application such as the increase of learning cost use threshold, professional specialists can unscramble result using analysis so on. According to these problems, this paper introduces a tag portrait technology which is based on includes general idea, structure key implementation The information system explained translated into language text in daily production concretized presentation research, help people pay attention...
With the continuous development of urbanization in China, number residents transformer area has continuously increased, which led to many problems management area. Among them, problem user relationship between meter and caused by change resident address is particularly serious, directly affects line loss assessment The power big data provides new solutions for above problems. Based on system data, this paper takes optimization algorithm as core establishes household verification model...
With the development of Internet Things and Artificial Intelligence, deploying inference tasks at network edge is widely adopted by operators so as to provide intelligent services. However, due limited resources relatively high resource consumption model, appropriate model ensure good accuracy can be challenging. In this work, we problem whose aim maximize overall efficiency benefit selecting AI models pre-trained on cloud servers deployed servers. a unable predict dynamic stochastic inputs...