A5000774921- Vibration and Dynamic Analysis
- Fire Detection and Safety Systems
- Belt Conveyor Systems Engineering
- Topic Modeling
- Multimodal Machine Learning Applications
- Power Line Inspection Robots
- Natural Language Processing Techniques
- Video Surveillance and Tracking Methods
Institute of Applied Physics and Computational Mathematics
2023-2024
Large language models perform well at common field, but are much less effective in niche academic fields .The cause of this problem is that large lack the ability to handle few-shot expert knowledge. To address issue, we purpose ProfLLM, a framework for adapting offline expert. ProfLLM provides knowledge processing capabilities through vector databases and prompt engineering, serializes Q-A pairs unstructured specialized associated completes parameter persistence by embedding fine-tuning,...
The traditional longitudinal tear detection methods have poor visible light imaging quality, generalization ability, and high hardware costs, then an improved encoder-decoder network was proposed to solve the problem. This method is a line structured for image acquisition system. Firstly, input images are downscaled by sorting algorithm; secondly, feature map semantically segmented reduced-dimensional network, pixel segmentation result mapped interval of tear; Finally, positioning length...