| ID | 原文 | 译文 |
| 7484 | 当前主流的方法是基于表示学习方法,通过神经网络对问题和答案进行向量表示,然后根据向量相似度对候选答案排序,该类方法忽略了问题和答案的局部关联性。 | The current mainstream method is based on the study method, through the neural network to questions and answers vectors, and then sorted according to the vector similarity to the candidate, the method ignores the local relevance of questions and answers. |
| 7485 | 针对这一问题,提出了一种基于多尺度相似度特征的深度学习模型。 | In order to solve this problem, this paper proposes a deep learning model based on multi-scale characteristics of similarity. |
| 7486 | 该模型采取传统的深度学习模型分别提取问题和答案的特征,然后计算各个尺度下的特征相似度得到问答的相似度矩阵,最后采取三种不同的相似度特征学习模型对相似度矩阵学习得到联合相似度。 | The depth of the model adopts the traditional learning model respectively to extract the characteristics of the questions and answers, and then calculate the characteristics of various scales similarity matrix of similarity for q&a, finally take similarity characteristics of three different learning model of learning to get joint similarity similarity matrix. |
| 7487 | 在公开数据集WebQA上进行实验验证,实验结果表明将相似度特征学习方法引入传统深度学习模型获得了较为明显的提升。 | WebQA public data sets on the experiment, the experimental results show that the similarity characteristics of learning methods into the depth of the traditional learning model obtained evident improvement. |
| 7488 | 偏置相位中心天线(displaced phase center antenna,DPCA)技术作为一种实现高分辨率、宽测绘带宽的有效技术手段,被广泛应用于高性能合成孔径雷达(synthetic aperture radar,SAR)系统研制中。 | Bias phase center antenna (displaced phase center can antenna, be used DPCA) technology as a kind of effective technologies to realize high resolution, wide bandwidth of surveying and mapping method, is widely used in high performance synthetic aperture radar (synthetic aperture radar, SAR) system is being developed. |
| 7489 | 为了实现最佳的成像效果,DPCA雷达成像系统需要满足方位向均匀采样条件,否则会导致方位向等效相位中心的周期性非均匀分布,进而使用经典匹配滤波成像方法会产生严重的方位模糊。 | In order to achieve the best imaging effect, DPCA radar imaging system need to satisfy the bearing to uniform sampling conditions, otherwise it will cause bearing to the non-uniform distribution of periodic equivalent phase center, and then use the classical matched filtering imaging method can produce serious bearing fuzzy. |
| 7490 | 根据稀疏信号处理理论,本文提出了一种基于复近似信息传递(complex approximate message passing,CAMP)算法的多通道非均匀采样DPCA成像方法,可以有效地解决多通道SAR因非均匀采样所产生的方位模糊以及杂波干扰问题,实现对观测区域的高精度重建。 | According to sparse signal processing theory, this paper puts forward a kind of based on complex approximate information transfer (complex approximate message passing, CAMP) algorithm for multi-channel non-uniform sampling DPCA imaging method, can effectively solve the multichannel SAR due to the non-uniform sampling bearing produced by fuzzy and clutter interference problems, achieve high precision of observation area reconstruction. |
| 7491 | 仿真和实际数据实验验证了该方法的有效性。 | The simulation and real data experiments verify the effectiveness of the proposed method. |
| 7492 | 针对水下目标特征提取问题,在卷积神经网的基础上,提出了一种新的网络结构。 | For feature extraction of underwater target, on the basis of the convolutional neural network, this paper proposes a new network structure. |
| 7493 | 该框架通过引入特征图多维加权层,强化了特征图的空间信息,弥补了进入全连接层时空间特征的丢失。 | The framework by introducing figure multi-dimension weighted layer, strengthen the characteristic figure of spatial information, make up into the space characteristics of the lost when the connection layer. |