1mlpack_local_coordinate_coding(1)User Commandsmlpack_local_coordinate_coding(1)
2
3
4

NAME

6       mlpack_local_coordinate_coding - local coordinate coding
7

SYNOPSIS

9        mlpack_local_coordinate_coding [-k int] [-i string] [-m unknown] [-l double] [-n int] [-N bool] [-s int] [-T string] [-o double] [-t string] [-V bool] [-c string] [-d string] [-M unknown] [-h -v]
10

DESCRIPTION

12       An  implementation  of  Local Coordinate Coding (LCC), which codes data
13       that approximately lives on a manifold using  a  variation  of  l1-norm
14       regularized  sparse  coding.  Given a dense data matrix X with n points
15       and d dimensions, LCC seeks to find a dense dictionary matrix D with  k
16       atoms  in d dimensions, and a coding matrix Z with n points in k dimen‐
17       sions. Because of the regularization method used, the atoms in D should
18       lie close to the manifold on which the data points lie.
19
20       The  original  data matrix X can then be reconstructed as D * Z. There‐
21       fore, this program finds a representation of  each  point  in  X  as  a
22       sparse linear combination of atoms in the dictionary D.
23
24       The  coding  is found with an algorithm which alternates between a dic‐
25       tionary step, which updates the dictionary D, and a coding step,  which
26       updates the coding matrix Z.
27
28       To  run  this  program, the input matrix X must be specified (with -i),
29       along with the number of atoms in the dictionary (-k). An initial  dic‐
30       tionary may also be specified with the '--initial_dictionary_file (-i)'
31       parameter. The l1-norm regularization parameter is specified  with  the
32       '--lambda  (-l)'  parameter.  For  example,  to  run LCC on the dataset
33       'data.csv' using 200 atoms and an l1-regularization parameter  of  0.1,
34       saving  the  dictionary  ’--dictionary_file  (-d)'  and  the codes into
35       '--codes_file (-c)', use
36
37       $ local_coordinate_coding --training_file data.csv --atoms 200 --lambda
38       0.1 --dictionary_file dict.csv --codes_file codes.csv
39
40       The maximum number of iterations may be specified with the '--max_iter‐
41       ations (-n)' parameter. Optionally, the input data matrix X can be nor‐
42       malized before coding with the '--normalize (-N)' parameter.
43
44       An  LCC  model may be saved using the '--output_model_file (-M)' output
45       parameter. Then, to encode new points  from  the  dataset  'points.csv'
46       with  the  previously saved model 'lcc_model.bin', saving the new codes
47       to ’new_codes.csv', the following command can be used:
48
49       $ local_coordinate_coding --input_model_file lcc_model.bin  --test_file
50       points.csv --codes_file new_codes.csv
51

OPTIONAL INPUT OPTIONS

53       --atoms (-k) [int]
54              Number of atoms in the dictionary. Default value 0.
55
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_dictionary_file (-i) [string]
63              Optional initial dictionary. Default value ''.
64
65       --input_model_file (-m) [unknown]
66              Input LCC model. Default value ''.
67
68       --lambda (-l) [double]
69              Weighted l1-norm regularization parameter.  Default value 0.
70
71       --max_iterations (-n) [int]
72              Maximum  number  of  iterations  for LCC (0 indicates no limit).
73              Default value 0.
74
75       --normalize (-N) [bool]
76              If set, the input data matrix will be normalized before coding.
77
78       --seed (-s) [int]
79              Random seed. If 0, 'std::time(NULL)' is used.  Default value 0.
80
81       --test_file (-T) [string]
82              Test points to encode. Default value ''.
83
84       --tolerance (-o) [double]
85              Tolerance for objective function. Default value 0.01.
86
87       --training_file (-t) [string]
88              Matrix of training data (X). Default value ''.
89
90       --verbose (-v) [bool]
91              Display informational messages and the full list  of  parameters
92              and timers at the end of execution.
93
94       --version (-V) [bool]
95              Display the version of mlpack.
96

OPTIONAL OUTPUT OPTIONS

98       --codes_file (-c) [string]
99              Output codes matrix. Default value ''.
100
101       --dictionary_file (-d) [string]
102              Output dictionary matrix. Default value ''.
103
104       --output_model_file (-M) [unknown]
105              Output for trained LCC model. Default value ''.
106

ADDITIONAL INFORMATION

108       For further information, including relevant papers, citations, and the‐
109       ory,  consult  the  documentation  found  at  http://www.mlpack.org  or
110       included with your distribution of mlpack.
111
112
113
114mlpack-3.0.4                   21 February 201m9lpack_local_coordinate_coding(1)