edu.stanford.nlp.rnn

Class RNNUtils

java.lang.Object

edu.stanford.nlp.rnn.RNNUtils

public class RNNUtils
extends java.lang.Object

Includes a bunch of utility methods usable by projects which use RNN, such as the parser and sentiment models. Some methods convert iterators of SimpleMatrix objects to and from a vector. Others are general utility methods on SimpleMatrix objects.

Author:

John Bauer, Richard Socher

Method Summary

Methods

Modifier and Type	Method and Description
static SimpleMatrix	concatenate(SimpleMatrix... vectors) Concatenates several column vectors into one large column vector
static SimpleMatrix	concatenateWithBias(SimpleMatrix... vectors) Concatenates several column vectors into one large column vector, adds a 1.0 at the end as a bias term
static SimpleMatrix	elementwiseApplyLog(SimpleMatrix input) Applies log to each of the entries in the matrix.
static SimpleMatrix	elementwiseApplyTanh(SimpleMatrix input) Applies tanh to each of the entries in the matrix.
static SimpleMatrix	elementwiseApplyTanhDerivative(SimpleMatrix input) Applies the derivative of tanh to each of the elements in the vector.
static double[]	paramsToVector(double scale, int totalSize, java.util.Iterator<SimpleMatrix>... matrices) Given a sequence of iterators over the matrices, builds a vector out of those matrices in the order given.
static double[]	paramsToVector(int totalSize, java.util.Iterator<SimpleMatrix>... matrices) Given a sequence of iterators over the matrices, builds a vector out of those matrices in the order given.
static SimpleMatrix	randomGaussian(int numRows, int numCols, java.util.Random rand) Returns a vector with random Gaussian values, mean 0, std 1
static java.util.Map<java.lang.String,SimpleMatrix>	readRawWordVectors(java.lang.String filename, int expectedSize) This method reads a file of raw word vectors, with a given expected size, and returns a map of word to vector.
static double	sigmoid(double x) Returns a sigmoid applied to the input x.
static SimpleMatrix	softmax(SimpleMatrix input) Applies softmax to all of the elements of the matrix.
static void	vectorToParams(double[] theta, java.util.Iterator<SimpleMatrix>... matrices) Given a sequence of Iterators over SimpleMatrix, fill in all of the matrices with the entries in the theta vector.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

vectorToParams

public static void vectorToParams(double[] theta,
                  java.util.Iterator<SimpleMatrix>... matrices)

Given a sequence of Iterators over SimpleMatrix, fill in all of the matrices with the entries in the theta vector. Errors are thrown if the theta vector does not exactly fill the matrices.

paramsToVector

public static double[] paramsToVector(int totalSize,
                      java.util.Iterator<SimpleMatrix>... matrices)

Given a sequence of iterators over the matrices, builds a vector out of those matrices in the order given. Asks for an expected total size as a time savings. AssertionError thrown if the vector sizes do not exactly match.

paramsToVector

public static double[] paramsToVector(double scale,
                      int totalSize,
                      java.util.Iterator<SimpleMatrix>... matrices)

Given a sequence of iterators over the matrices, builds a vector out of those matrices in the order given. The vector is scaled according to the scale parameter. Asks for an expected total size as a time savings. AssertionError thrown if the vector sizes do not exactly match.

readRawWordVectors

public static java.util.Map<java.lang.String,SimpleMatrix> readRawWordVectors(java.lang.String filename,
                                                              int expectedSize)

This method reads a file of raw word vectors, with a given expected size, and returns a map of word to vector.
The file should be in the format
WORD X1 X2 X3 ...
If vectors in the file are smaller than expectedSize, an exception is thrown. If vectors are larger, the vectors are trunccated and a warning is printed.

sigmoid

public static double sigmoid(double x)

Returns a sigmoid applied to the input x.

softmax

public static SimpleMatrix softmax(SimpleMatrix input)

Applies softmax to all of the elements of the matrix. The return matrix will have all of its elements sum to 1. If your matrix is not already a vector, be sure this is what you actually want.

elementwiseApplyLog

public static SimpleMatrix elementwiseApplyLog(SimpleMatrix input)

Applies log to each of the entries in the matrix. Returns a new matrix.

elementwiseApplyTanh

public static SimpleMatrix elementwiseApplyTanh(SimpleMatrix input)

Applies tanh to each of the entries in the matrix. Returns a new matrix.

elementwiseApplyTanhDerivative

public static SimpleMatrix elementwiseApplyTanhDerivative(SimpleMatrix input)

Applies the derivative of tanh to each of the elements in the vector. Returns a new matrix.

concatenateWithBias

public static SimpleMatrix concatenateWithBias(SimpleMatrix... vectors)

Concatenates several column vectors into one large column vector, adds a 1.0 at the end as a bias term

concatenate

public static SimpleMatrix concatenate(SimpleMatrix... vectors)

Concatenates several column vectors into one large column vector

randomGaussian

public static SimpleMatrix randomGaussian(int numRows,
                          int numCols,
                          java.util.Random rand)

Returns a vector with random Gaussian values, mean 0, std 1