void MyAggregator::Aggregate(const MyAggregator& aggregator) {
assert(aggregator.BinCount() == BinCount());
for (int b = 0; b < BinCount(); b++)
bins_[b] += aggregator.bins_[b];
sampleCount_ += aggregator.sampleCount_;
}
void HistogramAggregator::Aggregate(const HistogramAggregator& aggregator)
{
assert(aggregator.BinCount() == BinCount());
for (unsigned int b = 0; b < BinCount(); b++)
bins_[b] += aggregator.bins_[b];
sampleCount_ += aggregator.sampleCount_;
}
MyAggregator MyAggregator::DeepClone() const {
MyAggregator result(BinCount());
for (int b = 0; b < BinCount(); b++)
result.bins_[b] = bins_[b];
result.sampleCount_ = sampleCount_;
return result;
}
HistogramAggregator HistogramAggregator::DeepClone() const
{
HistogramAggregator result(BinCount());
for (unsigned int b = 0; b < BinCount(); b++)
result.bins_[b] = bins_[b];
result.sampleCount_ = sampleCount_;
return result;
}
//////////// IStatisticsAggregator implementation ////////////////
void HistogramAggregator::Clear()
{
for (unsigned int b = 0; b < BinCount(); b++)
bins_[b] = 0;
sampleCount_ = 0;
}
int MyAggregator::FindTallestBinIndex() const {
unsigned int maxCount = bins_[0];
int tallestBinIndex = 0;
for (int i = 1; i < BinCount(); i++) {
if (bins_[i] > maxCount) {
maxCount = bins_[i];
tallestBinIndex = i;
}
}
return tallestBinIndex;
}
double MyAggregator::Entropy() const {
if (sampleCount_ == 0)
return 0.0;
double result = 0.0;
for (int b = 0; b < BinCount(); b++)
{
double p = (double)bins_[b] / (double)sampleCount_;
result -= p == 0.0 ? 0.0 : p * log(p)/log(2.0);
}
return result;
}
bool
MaxLikelihoodSmoothing::Estimate(const ParamVector ¶ms,
const NgramLMMask *pMask,
ProbVector &probs,
ProbVector &bows) {
if (!_estimated) {
const CountVector &counts(_pLM->counts(_order));
const IndexVector &hists(_pLM->hists(_order));
// Compute inverse of sum of adjusted counts for each history.
CountVector histCounts(_pLM->sizes(_order - 1), 0);
ProbVector invHistCounts(histCounts.length());
BinCount(hists, histCounts);
invHistCounts = 1.0 / asDouble(histCounts);
// Compute maximum likelihood probability. 0 backoff.
probs = counts * invHistCounts[hists];
bows.set(0);
_estimated = true;
}
return true;
}
// IStatisticsAggregator implementation
void MyAggregator::Clear() {
for (int b = 0; b < BinCount(); b++)
bins_[b] = 0;
sampleCount_ = 0;
}
