首页>
外国专利>
Semantic Relationships Learning Facility, Semantic Relationships Learning Procedure, and Semantic Relationships Learning Program
Semantic Relationships Learning Facility, Semantic Relationships Learning Procedure, and Semantic Relationships Learning Program
展开▼
机译:语义关系学习设施,语义关系学习过程,以及语义关系学习计划
展开▼
页面导航
摘要
著录项
相似文献
摘要
A semantic relationship learning device (1) comprises a negative example data generation unit (12) which generates a plurality of negative example data pairs by combining speech data contained in a plurality of positive example data pairs, a feature value extraction unit (13) which generates feature values from extracts the speech data constituting each of the negative example data pairs, a similarity level calculating unit (14) which calculates a similarity level between the feature values in each of the negative example data pairs, a learning purpose negative example data generating unit (15) which converts the plurality of negative example data pairs into a plurality of predetermined ones Classifies similarity level areas on the basis of the similarity level, thereby generating a plurality of learning purpose negative example data corresponding to the plurality of similarity level areas, a learning data set selecting unit (17) which has learning purpose negative example data n set from the plurality of learning purpose negative example data sets in an order according to a selection plan which is predetermined based on the plurality of similarity level ranges, and a learning processing unit (18) that a machine learning process by using the selected learning purpose negative example data set and the Performs a variety of positive sample data pairs.
展开▼
机译:语义关系学习设备(1)包括否定示例数据生成单元(12),其通过组合包含在多个正示例数据对中的语音数据,特征值提取单元(13)来生成多个负示例数据对。从提取构成负示例数据对中的每一个的语音数据,相似度级计算单元(14)的语音数据生成特征值,该相似度级计算单元(14)计算每个负示例数据对中的每个否定示例数据对中的特征值之间的相似性等级,学习用途负面示例数据生成将多个负示例数据对转换为多个预定的单元(15)基于相似度水平对相似度级别区域进行分类,从而生成与多个相似级别区域对应的多个学习用途逆示例数据,一个学习数据集选择单元(17),其具有学习目的负极示例数据N SET F.根据多个相似度范围预先确定的选择计划,多个学习用途的多个学习目的负极示例数据集,以及通过使用所选择的学习目的否定的机器学习过程的学习处理单元(18)。示例数据集和执行各种正示例数据对。
展开▼