tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[doc] documentation fix for averaged perceptron

Browse Source
This commit is contained in:
Al
2015-09-08 16:37:23 -07:00
parent c80d8b8067
commit 607a607b71
1 changed files with 6 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
src/averaged_perceptron.h
View File

@@ -9,16 +9,17 @@ The averaged perceptron is a linear model, meaning the score for a given class
is the dot product of weights and the feature values.
This implementation of the averaged perceptron uses a trie data structure to
store the mapping from features to ids, which can be quite memory efficient
as opposed to a hash table and allows us to store
store the mapping from features to indices, which can be quite memory efficient
as opposed to a hash table and allows us to store millions of features with
very little memory.
The weights are stored as a sparse matrix in compressed sparse row format
(see sparse_matrix.h)
See [Collins, 2002] Discriminative Training Methods for Hidden Markov Models:
Theory and Experiments with Perceptron Algorithms
Paper: [Collins, 2002] Discriminative Training Methods for Hidden Markov Models:
Theory and Experiments with Perceptron Algorithms
Paper: http://www.cs.columbia.edu/~mcollins/papers/tagperc.pdf
Link: http://www.cs.columbia.edu/~mcollins/papers/tagperc.pdf
*/
#ifndef AVERAGED_PERCEPTRON_H
#define AVERAGED_PERCEPTRON_H