Chinese Journal of Liquid Crystals and Displays, Volume. 38, Issue 11, 1531(2023)
Aerospace information acquisition and image generation based on supervised contrastive learning
Figures & Tables(9)
Fig. 3. Flowchart of open source aerospace information acquisition and image generation framework
Table 1. Experimental environment configuration
View tableView in ArticleTable 1. Experimental environment configuration
实验工具 配置 CPU Intel（R）Xeon（R）Silver 4210R GPU NVIDIA GeForce RTX 3090 内存 128G 操作系统 Ubuntu 20.04.3 LTS 开发语言 Python 3.6
Table 2. Classification effect of aerospace category text
View tableView in ArticleTable 2. Classification effect of aerospace category text
模型 Precision Recall F1-score CNN 0.89 0.94 0.91 BiLSTM 0.72 0.80 0.76 Transformer 0.86 0.88 0.87 BiLSTM-Attention 0.94 0.98 0.96 BiLSTM-Attention+R-Drop 0.97 0.98 0.97
Table 3. Overall classification performance of each model
View tableView in ArticleTable 3. Overall classification performance of each model
模型 Precision Recall F1-score 预测时间/s CNN 0.88 0.90 0.88 11.96 CNN+R-Drop 0.89 0.90 0.88 11.20 BiLSTM 0.77 0.79 0.78 35.78 BiLSTM+R-Drop 0.79 0.80 0.80 37.52 Transformer 0.86 0.86 0.84 137.93 Transformer+R-Drop 0.87 0.88 0.87 138.78 BiLSTM-Attention 0.92 0.92 0.92 42.63 BiLSTM-Attention+R-Drop 0.93 0.93 0.93 41.16
Tools
Get Citation
Copy Citation Text
Yi-chen QI, Wei-chao ZHAO. Aerospace information acquisition and image generation based on supervised contrastive learning[J]. Chinese Journal of Liquid Crystals and Displays, 2023, 38(11): 1531
Paper Information
Category: Research Articles
Received: Feb. 15, 2023
Accepted: --
Published Online: Nov. 29, 2023
The Author Email: Wei-chao ZHAO (zhaoweichao@ciomp.ac.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20