ID 原文 译文
22295 在中文文本分类任务中，针对重要特征在中文文本中位置分布分散、稀疏的问题，以及不同文本特征对文本类别识别贡献不同的问题，该文提出一种基于语义理解的注意力神经网络、长短期记忆网络(LSTM)与卷积神经网络(CNN)的多元特征融合中文文本分类模型(3CLA)。 In Chinese text categorization tasks, the locations of the important features in the Chinese texts are disperse and sparse, and the different characteristics of Chinese texts contributes differently for the recognition of their categories. In order to solve the above problems, this paper proposes a multi-feature fusion model Three Convolutional neural network paths and Long short term memory path fused with Attention neural network path (3CLA) for Chinese text categorization, which is based on Convolutional Neural Network (CNN), Long Short Term Memory (LSTM) and semantic understanding attention neural networks.
22296 模型首先通过文本预处理将中文文本分词、向量化。 The model first uses text preprocessing to finish the segmentation and vectorization of the Chinese text.
22297 然后，通过嵌入层分别经过 CNN 通路、LSTM 通路和注意力算法模型通路以提取不同层次、具有不同特点的文本特征。 Then, through the embedding layer, the input data are sent to the CNN path, the LSTM path and the attention path respectively to extract text features of different levels and different characteristics.
22298 最终，文本特征经融合层融合后，由 softmax 分类器进行分类。 Finally, the text features are fused by the fusion layer and classified by the classifier.
22299 基于中文语料进行了文本分类实验。 Based on the Chinese corpus, the text classification experiment is carried out.
22300 实验结果表明，相较于 CNN 结构模型与 LSTM 结构模型，提出的算法模型对中文文本类别的识别能力最多提升约 8%。 The results of the experiments show that compared with the CNN structure model and the LSTM structure model, the proposed algorithm model improves the recognition ability of Chinese text categories by up to about 8%.
22301 研究学者们认为指数加权均值比(ROEWA)算子存在无法计算 SAR 图像边缘方向的缺陷。 Researchers generally consider that Ratio Of Exponentially Weighted Averages (ROEWA) can not calculate the edge directions of SAR images.
22302 为此，进行了一些通过方向滤波器为 ROEWA 算法施加方向的工作。 Therefore, some directional filters are used to add directions to ROEWA.
22303 该文对 ROEWA 算法进行了深入的探讨和分析，通过对ROEWA 算法卷积过程的进一步推导，获得了 ROEWA 算法像素级的观测公式。 In this paper, an Enhanced ROEWA (EROEWA) algorithm is proposed. Through the further derivation of the ROEWA algorithm convolution process, the pixel-level observation formula of ROEWA algorithm is obtained.
22304 首先，利用新的卷积策略将 ROEWA 的公式项解耦，获得了 4 个方向的指数加权均值； First, a new convolution strategy is used to decouple the ROEWA formula to obtain exponentially weighted averages over the four directions.