Invention Grant
US08386400B2 Unsupervised clustering of multimedia data using a large-scale matching system 有权
使用大型匹配系统进行多媒体数据的无监督聚类

Unsupervised clustering of multimedia data using a large-scale matching system
Abstract:
A method and apparatus for clustering a plurality of data elements. The method comprises receiving a plurality of cluster elements, each cluster element containing at least a data element; generating a clustering score for each cluster element of the plurality of cluster elements versus all other cluster elements of the plurality of cluster elements using a computing device; determining a size of a diagonal matrix having a size corresponding to the number of the plurality of cluster elements; placing the clustering score in a diagonal matrix in storage one clustering score for each pair of cluster elements; creating a new cluster element for each two cluster elements in the diagonal matrix having a clustering score that exceeds a threshold; and storing generated new cluster elements in the storage.
Information query
Patent Agency Ranking
0/0