ID 原文 译文
56858 其次,由于当前任务的变化,会有一些新词出现,这些新词的词向量不能通过预训练的词向量获得,因此本文提出了一种保持语义关系的词向量复用算法(SrpWer). In addition, ascorpus topics change, new words will appear for a given task, and their corresponding embeddings cannot beobtained from pretrained vectors. Therefore, to reuse word embeddings, we propose a semantic relation preservedword embedding reuse method.
56859 SrpWer首先对当前数据集中词语之间的关系进行建模,然后结合预训练的词向量生成新词对应的词向量. The proposed method first learns word relations from the current corpus. Then,pretrained word embeddings are utilized to help generate embeddings for new observed words.
56860 实验结果验证了SrpWer的有效性. Experimentalresults verify the effectiveness of the proposed method.
56861 集成特征选择算法将多种特征选择方法结果结合在一起,旨在得到更加有效的特征子集. Ensemble feature selection algorithms aggregate the results of multiple feature selection methodsin order to select an effective subset of features.
56862 然而这些算法通常假设每种特征选择方法是平等的,没有考虑不同特征选择方法性能的差异性,导致少数方法选择出的有效特征被忽略. However, typically, ensemble algorithms treat each featureselection method equally and do not consider performance differences. Consequently, features selected by arelatively smaller number of methods may not be included.
56863 为解决这一问题,本文提出一种可以有效地结合不同特征选择方法优势,并利用专家的知识逐步改善所选特征的交互式特征选择方法. To address this problem, we propose an interactivefeature selection method that can more effectively aggregate the results of multiple feature selection methodsand iteratively improve the selected features by integrating expert knowledge.
56864 该方法包括一个基于众包学习的集成特征选择算法和一个基于该算法开发的可视分析系统. The proposed method includesa learning-from-crowds-based ensemble feature selection algorithm and a visual analysis system.
56865 基于众包学习的集成特征选择算法利用众包学习模型对不同特征选择方法的性能进行建模,计算每种方法的可靠性,并在此基础上将这些方法的结果有机融合. The algorithmmodels the performance of multiple feature selection methods, calculates their reliabilities, and aggregates results.
56866 可视分析系统提供了丰富的排序方式,帮助专家理解单个特征选择方法的特征选择结果和特征在分类任务中所起的作用,从而让专家交互迭代地改善现有特征子集. To integrate expert knowledge, the visual analysis system provides a set of ranking schemes to assist experts inunderstanding the results of an individual feature selection method and the roles played by the features inclassification tasks.
56867 在4个真实世界数据集上的数值实验表明,相比于现有的集成特征选择算法,本文提出的算法能够带来0.63%～2. 85%分类准确率的提升. A numerical experiment conducted on four real-world datasets shows that the proposedalgorithm can improve classification accuracy by 0. 63%–2. 85% compared to state-of-the-art ensemble featureselection algorithms.