/**
* This is the method to call for assigning labels and node vectors to the Tree. After calling
* this, each of the non-leaf nodes will have the node vector and the predictions of their classes
* assigned to that subtree's node.
*/
public void forwardPropagateTree(Tree tree) {
INDArray nodeVector;
INDArray classification;
if (tree.isLeaf()) {
// We do nothing for the leaves. The preterminals will
// calculate the classification for this word/tag. In fact, the
// recursion should not have gotten here (unless there are
// degenerate trees of just one leaf)
throw new AssertionError("We should not have reached leaves in forwardPropagate");
} else if (tree.isPreTerminal()) {
classification = getUnaryClassification(tree.label());
String word = tree.children().get(0).value();
INDArray wordVector = getFeatureVector(word);
if (wordVector == null) {
wordVector = featureVectors.vector(Word2Vec.UNK);
}
nodeVector =
Nd4j.getExecutioner()
.execAndReturn(Nd4j.getOpFactory().createTransform(activationFunction, wordVector));
} else if (tree.children().size() == 1) {
throw new AssertionError(
"Non-preterminal nodes of size 1 should have already been collapsed");
} else if (tree.children().size() == 2) {
Tree left = tree.firstChild(), right = tree.lastChild();
forwardPropagateTree(left);
forwardPropagateTree(right);
String leftCategory = tree.children().get(0).label();
String rightCategory = tree.children().get(1).label();
INDArray W = getBinaryTransform(leftCategory, rightCategory);
classification = getBinaryClassification(leftCategory, rightCategory);
INDArray leftVector = tree.children().get(0).vector();
INDArray rightVector = tree.children().get(1).vector();
INDArray childrenVector = Nd4j.appendBias(leftVector, rightVector);
if (useDoubleTensors) {
INDArray doubleT = getBinaryINDArray(leftCategory, rightCategory);
INDArray INDArrayIn = Nd4j.concat(0, leftVector, rightVector);
INDArray INDArrayOut = Nd4j.bilinearProducts(doubleT, INDArrayIn);
nodeVector =
Nd4j.getExecutioner()
.execAndReturn(
Nd4j.getOpFactory()
.createTransform(
activationFunction, W.mmul(childrenVector).addi(INDArrayOut)));
} else
nodeVector =
Nd4j.getExecutioner()
.execAndReturn(
Nd4j.getOpFactory()
.createTransform(activationFunction, W.mmul(childrenVector)));
} else {
throw new AssertionError("Tree not correctly binarized");
}
INDArray inputWithBias = Nd4j.appendBias(nodeVector);
if (inputWithBias.rows() != classification.columns()) inputWithBias = inputWithBias.transpose();
INDArray preAct = classification.mmul(inputWithBias);
INDArray predictions =
Nd4j.getExecutioner()
.execAndReturn(Nd4j.getOpFactory().createTransform(outputActivation, preAct));
tree.setPrediction(predictions);
tree.setVector(nodeVector);
}