1mlpack_krann(1) General Commands Manual mlpack_krann(1)
2
6 mlpack_krann - k-rank-approximate-nearest-neighbors (krann)
7
9 mlpack_krann [-h] [-v]
10
12 This program will calculate the k rank-approximate-nearest-neighbors of
13 a set of points. You may specify a separate set of reference points and
14 query points, or just a reference set which will be used as both the
15 reference and query set. You must specify the rank approximation (in %)
16 (and optionally the success probability).
17 For example, the following will return 5 neighbors from the top 0.1% of
19 the data (with probability 0.95) for each point in 'input.csv' and
20 store the distances in 'distances.csv' and the neighbors in the file
21 'neighbors.csv':
22 $ allkrann -k 5 -r input.csv -d distances.csv -n neighbors.csv --tau
24 0.1
25 Note that tau must be set such that the number of points in the corre‐
27 sponding percentile of the data is greater than k. Thus, if we choose
28 tau = 0.1 with a dataset of 1000 points and k = 5, then we are attempt‐
29 ing to choose 5 nearest neighbors out of the closest 1 point -- this is
30 invalid and the program will terminate with an error message.
31 The output files are organized such that row i and column j in the
33 neighbors output file corresponds to the index of the point in the ref‐
34 erence set which is the i'th nearest neighbor from the point in the
35 query set with index j. Row i and column j in the distances output
36 file corresponds to the distance between those two points.
37
39 --alpha (-a) [double]
40 The desired success probability. Default value 0.95.
41 --first_leaf_exact (-X)
43 The flag to trigger sampling only after exactly exploring the
44 first leaf.
45 --help (-h)
47 Default help info.
48 --info [string]
50 Get help on a specific module or option. Default value ''.
51 --input_model_file (-m) [string] File containing pre-trained kNN
52 model. Default value ''.
53 --k (-k) [int]
55 Number of nearest neighbors to find. Default value 0.
56 --leaf_size (-l) [int]
58 Leaf size for tree building (used for kd-trees, UB trees, R
59 trees, R* trees, X trees, Hilbert R trees, R+ trees, R++ trees,
60 and octrees). Default value 20.
61 --naive (-N)
63 If true, sampling will be done without using a tree.
64 --query_file (-q) [string]
66 File containing query points (optional). Default value ''.
67 --random_basis (-R)
69 Before tree-building, project the data onto a random orthogonal
70 basis. --reference_file (-r) [string] File containing the ref‐
71 erence dataset. Default value ''.
72 --sample_at_leaves (-L)
74 The flag to trigger sampling at leaves.
75 --seed [int]
77 Random seed (if 0, std::time(NULL) is used). Default value 0.
78 --single_mode (-s)
80 If true, single-tree search is used (as opposed to dual-tree
81 search. --single_sample_limit (-S) [int] The limit on the maxi‐
82 mum number of samples (and hence the largest node you can
83 approximate). Default value 20.
84 --tau (-T) [double]
86 The allowed rank-error in terms of the percentile of the data.
87 Default value 5.
88 --tree_type (-t) [string]
90 Type of tree to use: 'kd', 'ub', 'cover', 'r', ’x', 'r-star',
91 'hilbert-r', 'r-plus', ’r-plus-plus', 'oct'. Default value 'kd'.
92 --verbose (-v)
94 Display informational messages and the full list of parameters
95 and timers at the end of execution.
96 --version (-V)
98 Display the version of mlpack.
99
101 --distances_file (-d) [string] File to output distances into. Default
102 value ’'. --neighbors_file (-n) [string] File to output neighbors
103 into. Default value ’'. --output_model_file (-M) [string] If speci‐
104 fied, the kNN model will be saved to the given file. Default value ''.
105
108 For further information, including relevant papers, citations, and the‐
109 ory, For further information, including relevant papers, citations, and
110 theory, consult the documentation found at http://www.mlpack.org or
111 included with your consult the documentation found at
112 http://www.mlpack.org or included with your DISTRIBUTION OF MLPACK.
113 DISTRIBUTION OF MLPACK.
114 mlpack_krann(1)