ID 原文 译文
18535 首先，使用余弦相似度构建标记相关性矩阵，通过谱聚类将标记分组以提取各标记组的类属属性，减少计算全部标记类属属性的时间消耗。 Firstly, the cosinesimilarity is used to construct the label correlation matrix, and the class labels are grouped by spectralclustering to extract the label-specific features of each label group to reduce the time consumption forcalculating the label-specific features of all class labels.
18536 然后，计算各标记密度以更新标记空间矩阵，将标记密度信息加入原标记中，扩大正负标记的间隔，通过标记密度分类间隔面的方法有效解决标记分布密度不平衡问题。 Then, the density of each label is calculated to updatethe label space matrix, the label-density information is added to the original label space. The classificationmargin between the positive and negative labels is expanded, thus the imbalance label distribution densityproblem is effectively solved by the method of label-density classification margin.
18537 最后，通过将组类属属性和标记密度矩阵输入极限学习机以得到最终分类模型。 Finally, the final classificationmodel is obtained by inputting the group-label-specific features and the label-density matrix into the extremelearning machine.
18538 对比实验充分验证了该文所提算法的可行性与稳定性。 The comparison experiment results verify fully the feasibility and stability of the proposed algorithm.
18539 特征子空间学习是图像识别及分类任务的关键技术之一，传统的特征子空间学习模型面临两个主要的问题。 Feature subspace learning is a critical technique in image recognition and classification tasks.Conventional feature subspace learning methods include two main problems.
18540 一方面是如何使样本在投影到特征空间后有效地保持其局部结构和判别性。 One is how to preserve the localstructures and discrimination when the samples are projected into the learned subspace.
18541 另一方面是当样本含噪时传统学习模型所发生的失效问题。 The other hand whenthe data are corrupted with noise, the conventional learning models usually do not work well.
18542 针对上述两个问题，该文提出一种基于低秩表示(LRR)的判别特征子空间学习模型，该模型的主要贡献包括： To solve the twoproblems, a discriminative feature learning method is proposed based on Low Rank Representation (LRR). Thenovel method includes three main contributions.
18543 通过低秩表示探究样本的局部结构，并利用表示系数作为样本在投影空间的相似性约束，使投影子空间能够更好地保持样本的局部近邻关系； It explores the local structures among samples via low rankrepresentation, and the representation coefficients are used as the similarity measurement to preserve the localneighborhood existed in the samples;
18544 为提高模型的抗噪能力，构造了一种利用低秩重构样本的判别特征学习约束项，同时增强模型的判别性和鲁棒性； To improve the anti-noise performance, a discriminative learning item is constructed from the recovered samples via low rank representation, which can enhance the discrimination and robustness simultaneously;