期刊文献+
共找到2篇文章
< 1 >
每页显示 20 50 100
基于多覆盖模型的神经机器翻译 被引量:8
1
作者 刘俊鹏 黄锴宇 +2 位作者 李玖一 宋鼎新 黄德根 《软件学报》 EI CSCD 北大核心 2022年第3期1141-1152,共12页
覆盖模型可以缓解神经机器翻译中的过度翻译和漏翻译问题.现有方法通常依靠覆盖向量或覆盖分数等单一方式存储覆盖信息,而未考虑不同覆盖信息之间的关联性,因此对信息的利用并不完善.针对该问题,基于翻译历史信息的一致性和模型之间的... 覆盖模型可以缓解神经机器翻译中的过度翻译和漏翻译问题.现有方法通常依靠覆盖向量或覆盖分数等单一方式存储覆盖信息,而未考虑不同覆盖信息之间的关联性,因此对信息的利用并不完善.针对该问题,基于翻译历史信息的一致性和模型之间的互补性,提出了多覆盖融合模型.首先定义词级覆盖分数概念;然后利用覆盖向量和覆盖分数存储的信息同时指导注意力机制,降低信息存储损失对注意力权重计算的影响.根据两种覆盖信息融合方式的不同,提出了两种多覆盖融合方法.利用序列到序列模型在中英翻译任务上进行了实验,结果表明,所提方法能够显著提升翻译性能,并改善源语言和目标语言的对齐质量.与只使用覆盖向量的模型相比,过度翻译和漏翻译问题的数量得到进一步减少. 展开更多
关键词 神经机器翻译 注意力机制 序列到序列模型 多覆盖模型 过度翻译 漏翻译
下载PDF
Human Settlement Analysis Based on Multi-temporal Remote Sensing Data: A Case Study of Xuzhou City, China 被引量:5
2
作者 ZHU Jishuai TIAN Shufang +1 位作者 TAN Kun DU Peijun 《Chinese Geographical Science》 SCIE CSCD 2016年第3期389-400,共12页
To evaluate urban human settlement, we propose a human settlement environment development index(HSEDI) model by choosing vegetation coverage, land surface temperature, impervious surfaces, slope, wetness, and water co... To evaluate urban human settlement, we propose a human settlement environment development index(HSEDI) model by choosing vegetation coverage, land surface temperature, impervious surfaces, slope, wetness, and water condition as the evaluation factors. We applied the proposed model to Xuzhou City, Jiangsu Province, China. Landsat-5 Thematic Mapper(TM) images from 1998 to 2010 and digital elevation model(DEM) data with a 30-m resolution were used to calculate the values of the six evaluation factors. The HSEDI value in Xuzhou City was found to be between 2.24 and 8.10 from 1998 to 2010, and it was further divided into five levels, uninhabitable, moderately uninhabitable, generally inhabitable, moderately inhabitable, and inhabitable. The best HSEDI value was in 2007. The generally inhabitable region was about 100.98 km^2, covering 30.87% of the total area in 2007; the moderately inhabitable region was about 170.58 km2 covering 52.15% of the total area; the inhabitable region was about 32.03 km^2, covering 9.79% of the total area; the percentage of the uninhabitable region was zero; and that of the moderately uninhabitable region was very small, less than 1.00%. Moreover, we analyzed the habitability in the respect of spatial patterns and change detection. Results show that the degraded regions of habitability quality are mainly located in the urban fringe and the improved regions are mainly located in the main urban and rural areas. Reason for the degraded habitability quality is the rapid progress of urbanization. However, the increase in urban green spaces and the construction of the main urban area promoted the improved habitability quality. Besides, we further analyzed socio-economic and socio-demographic data to confirm the results of the habitability analysis. The results indicate that the human settlement in Xuzhou City is in a satisfactory condition, but some efforts should be made to control the possible uninhabitable and moderately uninhabitable regions, and to improve the quality of the generally inhabitable regions. 展开更多
关键词 habitability human settlement Landsat human settlement environment development index(HSEDI) model
下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部