Research Article
Matrix Factorization for Evolution Data
Input: matrix ; number of the latent features ; learning rates , and ; | regularization parameters , and ; threshold | Output: the estimated matrix . | // Initialize and . | (1) Generate random vectors ; | (2) for ; ; do | (3) Generate with | (4) Let | (5) end | // Coordinate descent. | (6) ; | (7) ; | (8) while do | (9) ; | (10) for do | (11) Let ; | (12) end | (13) ; | (14) for do | (15) Let ; | (16) end | (17) Replace the all with , and with , recompute ; | (18) end | (19) return ; |
|