public int[] LearnFromOneInstance(int[][] wordsWithActiveFeatures, int[] labelledTags)
{
// averaged percetron code : http://ciml.info/dl/v0_8/ciml-v0_8-ch03.pdf
int numWords = wordsWithActiveFeatures.Length;
Viterbi.Decode(
wordsWithActiveFeatures,
w_alphaTagFeature,
w_alphaTagPreviousTag,
latticeBuffer,
backPointerBuffer,
numTags,
decodedTagsBuffer);
//find mistakes
int diff = 0;
for (int t = 0; t < numWords; t++)
{
if (decodedTagsBuffer[t] != labelledTags[t])
{
diff++;
}
}
loss += diff * 1.0 / numWords;
instances++;
if (c % showFactor == 0)
{
showFactor *= 2;
Console.WriteLine("Average loss {0} = {1}", c, loss / instances);
instances = 0;
loss = 0;
}
//do perceptron update
if (diff > 0)
{
for (int t = 0; t < numWords; t++)
{
int dt = decodedTagsBuffer[t]; //decoded tag
int lt = labelledTags[t]; //labelled tag
if (decodedTagsBuffer[t] != labelledTags[t])
{
//update W
foreach (int f in wordsWithActiveFeatures[t])
{
w_alphaTagFeature[dt, f] -= 1;
c_alphaTagFeature[dt, f] -= c;
w_alphaTagFeature[lt, f] += 1;
c_alphaTagFeature[lt, f] += c;
}
}
if (t > 0)
{
int dpt = decodedTagsBuffer[t - 1]; //decoded previous tag
int lpt = labelledTags[t - 1]; //labelled previous tag
if (dt != lt || dpt != lpt)
{
w_alphaTagPreviousTag[dt, dpt] -= 1;
c_alphaTagPreviousTag[dt, dpt] -= c;
w_alphaTagPreviousTag[lt, lpt] += 1;
c_alphaTagPreviousTag[lt, lpt] += c;
}
}
}
c += 1;
}
return(decodedTagsBuffer);
}
public void Label(int[][] wordsWithActiveFeatures, int[] tags)
{
Viterbi.Decode(wordsWithActiveFeatures, alphaTagFeature, alphaTagPreviousTag, latticeBuffer, backPointerBuffer, numTags, tags);
}
