Factorization Machines Study notes (i) Predictive tasks

     
    Recently learned an algorithm called Factorization Machines (FM), which can predict arbitrary real-valued vectors. The main advantages include: 1) can be used in highly sparse data scenarios, 2) with linear computational complexity. In this paper, the FM framework is briefly introduced, and its training algorithm-random gradient descent (SGD) method and alternating least squares (ALS) method are deduced in detail.

RELATED LINKS :

(i) Forecast tasks

(ii) Model equations

(iii) Return and classification

(iv) Learning algorithms

Author: peghoty

Source: http://blog.csdn.net/itplus/article/details/40534885

Welcome to reprint/share, but be sure to declare the source of the article.

Factorization Machines Study notes (i) Predictive tasks