1mlpack_nmf(1)                    User Commands                   mlpack_nmf(1)
2
3
4

NAME

6       mlpack_nmf - non-negative matrix factorization
7

SYNOPSIS

9        mlpack_nmf -i string -r int [-q string] [-p string] [-m int] [-e double] [-s int] [-u string] [-V bool] [-H string] [-W string] [-h -v]
10

DESCRIPTION

12       This  program  performs  non-negative matrix factorization on the given
13       dataset, storing the resulting decomposed  matrices  in  the  specified
14       files. For an input dataset V, NMF decomposes V into two matrices W and
15       H such that
16
17       V = W * H
18
19       where all elements in W and H are non-negative. If V is of  size  (n  x
20       m), then W will be of size (n x r) and H will be of size (r x m), where
21       r is the rank of the factorization  (specified  by  the  '--rank  (-r)'
22       parameter).
23
24       Optionally, the desired update rules for each NMF iteration can be cho‐
25       sen from the following list:
26
27              ·  multdist: multiplicative distance-based update rules (Lee and
28                 Seung 1999)
29
30              ·  multdiv:  multiplicative  divergence-based  update rules (Lee
31                 and Seung 1999)
32
33              ·  als: alternating least squares update rules (Paatero and Tap‐
34                 per 1994)
35
36       The  maximum  number  of iterations is specified with '--max_iterations
37       (-m)', and the minimum residue required for  algorithm  termination  is
38       specified with the '--min_residue (-e)' parameter.
39
40       For  example,  to  run NMF on the input matrix 'V.csv' using the 'mult‐
41       dist' update rules with a rank-10 decomposition and storing the  decom‐
42       posed matrices into 'W.csv' and 'H.csv', the following command could be
43       used:
44
45       $ nmf --input_file  V.csv  --w_file  W.csv  --h_file  H.csv  --rank  10
46       --update_rules multdist
47

REQUIRED INPUT OPTIONS

49       --input_file (-i) [string]
50              Input dataset to perform NMF on.
51
52       --rank (-r) [int]
53              Rank of the factorization.
54

OPTIONAL INPUT OPTIONS

56       --help (-h) [bool]
57              Default help info.
58
59       --info [string]
60              Get help on a specific module or option.  Default value ''.
61
62       --initial_h_file (-q) [string]
63              Initial H matrix. Default value ''.
64
65       --initial_w_file (-p) [string]
66              Initial W matrix. Default value ''.
67
68       --max_iterations (-m) [int]
69              Number of iterations before NMF terminates (0 runs until conver‐
70              gence. Default value 10000.
71
72       --min_residue (-e) [double]
73              The minimum root mean square residue allowed for each iteration,
74              below which the program terminates. Default value 1e-05.
75
76       --seed (-s) [int]
77              Random  seed. If 0, 'std::time(NULL)' is used.  Default value 0.
78              --update_rules (-u) [string] Update rules for each iteration;  (
79              multdist | multdiv | als ). Default value 'multdist'.
80
81       --verbose (-v) [bool]
82              Display  informational  messages and the full list of parameters
83              and timers at the end of execution.
84
85       --version (-V) [bool]
86              Display the version of mlpack.
87

OPTIONAL OUTPUT OPTIONS

89       --h_file (-H) [string]
90              Matrix to save the calculated H to. Default value ''.
91
92       --w_file (-W) [string]
93              Matrix to save the calculated W to. Default value ''.
94

ADDITIONAL INFORMATION

96       For further information, including relevant papers, citations, and the‐
97       ory,  consult  the  documentation  found  at  http://www.mlpack.org  or
98       included with your distribution of mlpack.
99
100
101
102mlpack-3.0.4                   21 February 2019                  mlpack_nmf(1)