Mercurial > hg > chime-home-dataset-annotation-and-baseline-evaluation-code
changeset 4:39258b875228
Disable sampling when estimating GMMs -- use entire set of training samples.
| author | peterf |
|---|---|
| date | Tue, 21 Jul 2015 14:16:58 +0100 |
| parents | 0147bf388eb8 |
| children | b523456082ca |
| files | gmm_baseline_experiments/figures/predictionperformanceGMM_Baseline_EXPER005.pdf gmm_baseline_experiments/gmm/GMM_methods.py |
| diffstat | 2 files changed, 6 insertions(+), 5 deletions(-) [+] |
line wrap: on
line diff
Binary file gmm_baseline_experiments/figures/predictionperformanceGMM_Baseline_EXPER005.pdf has changed
--- a/gmm_baseline_experiments/gmm/GMM_methods.py Sun Jul 19 22:11:19 2015 +0100 +++ b/gmm_baseline_experiments/gmm/GMM_methods.py Tue Jul 21 14:16:58 2015 +0100 @@ -33,7 +33,8 @@ # scale data [X,trainMean,trainStd] = scale(X) - numTrainingExamplesOfEachType = 20000 + #No sampling; use entire set of frames + #numTrainingExamplesOfEachType = 20000 X = X.tolist() Y = Y.tolist() @@ -48,15 +49,15 @@ #Sample data for label in Labels: DataByLabel[label] = np.array(DataByLabel[label]) - I = np.random.choice(DataByLabel[label].shape[0],numTrainingExamplesOfEachType) - DataByLabel[label] = DataByLabel[label][I] + #I = np.random.choice(DataByLabel[label].shape[0],numTrainingExamplesOfEachType) + #DataByLabel[label] = DataByLabel[label][I] - #print "Training..." + print "Training..." GMMS = {} for label in Labels: GMMS[label] = mixture.GMM(n_components=numComponents, covariance_type=covarianceType) GMMS[label].fit(DataByLabel[label]) - #print "Done!" + print "Done!" startTime = time.time()