Non-negative Matrix Factorization Algorithm Based on Method of Least Square

摘要：

The nonnegative matrices decomposition algorithm is discussed, and its objective function based on Euclidean distance proposed by Lee & Seung is simplified. A decomposition factor is identified in the primitive matrix at firstly and the other factor is solved using least-square method. And then, using least-square method again to solve another factor for the obtained result. So circulates, until the solution with minimum error will be found. The specific algorithm is described and it is easily to be realized by using LINGO software. An example shows that the matrix decomposition result has smaller error than the result of existing.

关键词： non-negative matrix factorization method of least square Euclidean distance LINGO software

作者: Gao Zunhai Xu Taiyan

作者单位: Dept.of Mathematics and physics Wuhan Polytechnic University Wuhan, Hubei, China

会议类型: 国际会议

会议名称: The 2010 International Conference on Computer Application and System Modeling(2010计算机应用与系统建模国际会议 ICCASM 2010)

会议地点: 太原

会议语种:英文

页码: 512-515

在线出版日期: 2010-10-22（万方平台首次上网日期，不代表论文的发表时间）