Taken:
- ❏ UD-info opnemen in XML
- ❏ Bestaande corpora bijwerken
❏ Visualisatie zonder JavaScript-library
- ✔ Weergave van gewone UD en extended UD
- ❏ nauwkeuriger tekstbreedte: webfonts?
- ❏ tooltip verbeteren (kleuren, positie)
- ❏ als te breed voor venster, dan in stukken verdelen?
- ❏ Downloaden van UD, per zin en voor een heel corpus
- ❏ Beschrijving op info-pagina
- ❏ UD-info opnemen in XML
✔ Proef in ~/spraak/fa/mfcc herhalen, nu met mfcc per tiende(bijv) woord i.p.v. per klank
- → zie beneden
Let op: Alle waardes voor O/Z Baseline zijn fout
missing = 0 een sample per klank N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.531 0.819 0.757 0.529 0.764 SVM: linear 0.434 0.620 0.628 0.527 0.653 SVM: rbf 0.646 0.948 0.951 0.541 0.909 AdaBoost 0.379 0.623 0.564 0.505 0.555 Gaussian Naive Bayes missing = NaN een sample per klank N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.655 0.902 0.887 0.557 0.873 SVM: linear 0.395 0.611 0.527 0.493 0.601 SVM: rbf 0.619 0.931 0.938 0.559 0.900 AdaBoost 0.385 0.615 0.551 0.489 0.560 Gaussian Naive Bayes
missing = 0 samples per lettergreep = 2 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.649 0.899 0.874 0.574 0.842 SVM: linear 0.411 0.629 0.557 0.520 0.584 SVM: rbf 0.643 0.913 0.918 0.514 0.907 AdaBoost 0.366 0.598 0.554 0.454 0.551 Gaussian Naive Bayes missing = NaN samples per lettergreep = 2 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.654 0.906 0.889 0.573 0.865 SVM: linear 0.355 0.598 0.516 0.420 0.548 SVM: rbf 0.623 0.919 0.915 0.574 0.915 AdaBoost 0.381 0.595 0.557 0.485 0.561 Gaussian Naive Bayes missing = 0 samples per lettergreep = 3 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.620 0.873 0.868 0.572 0.818 SVM: linear 0.404 0.611 0.544 0.509 0.627 SVM: rbf 0.662 0.917 0.920 0.565 0.926 AdaBoost 0.375 0.596 0.562 0.434 0.561 Gaussian Naive Bayes missing = NaN samples per lettergreep = 3 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.670 0.906 0.894 0.603 0.862 SVM: linear 0.382 0.560 0.536 0.461 0.597 SVM: rbf 0.680 0.944 0.928 0.514 0.924 AdaBoost 0.393 0.596 0.562 0.470 0.558 Gaussian Naive Bayes missing = 0 samples per lettergreep = 4 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.612 0.891 0.837 0.573 0.820 SVM: linear 0.437 0.627 0.609 0.521 0.611 SVM: rbf 0.638 0.920 0.927 0.568 0.912 AdaBoost 0.391 0.592 0.570 0.500 0.556 Gaussian Naive Bayes missing = NaN samples per lettergreep = 4 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.682 0.902 0.874 0.627 0.871 SVM: linear 0.413 0.594 0.593 0.506 0.561 SVM: rbf 0.639 0.938 0.899 0.548 0.899 AdaBoost 0.375 0.583 0.559 0.473 0.548 Gaussian Naive Bayes missing = 0 samples per lettergreep = 5 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.620 0.875 0.837 0.590 0.825 SVM: linear 0.420 0.641 0.558 0.544 0.629 SVM: rbf 0.635 0.940 0.936 0.557 0.895 AdaBoost 0.365 0.594 0.554 0.454 0.545 Gaussian Naive Bayes missing = NaN samples per lettergreep = 5 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.660 0.911 0.892 0.595 0.852 SVM: linear 0.408 0.632 0.562 0.492 0.598 SVM: rbf 0.625 0.920 0.933 0.620 0.908 AdaBoost 0.387 0.596 0.561 0.502 0.550 Gaussian Naive Bayes missing = 0 samples per lettergreep = 6 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.597 0.895 0.838 0.552 0.814 SVM: linear 0.406 0.621 0.557 0.546 0.621 SVM: rbf 0.619 0.944 0.936 0.545 0.910 AdaBoost 0.377 0.576 0.543 0.507 0.542 Gaussian Naive Bayes missing = NaN samples per lettergreep = 6 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.643 0.897 0.886 0.572 0.854 SVM: linear 0.383 0.618 0.565 0.468 0.612 SVM: rbf 0.630 0.933 0.917 0.566 0.897 AdaBoost 0.379 0.587 0.550 0.494 0.540 Gaussian Naive Bayes missing = 0 samples per lettergreep = 7 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.572 0.851 0.837 0.506 0.782 SVM: linear 0.425 0.619 0.618 0.471 0.606 SVM: rbf 0.611 0.911 0.920 0.609 0.907 AdaBoost 0.390 0.603 0.551 0.491 0.566 Gaussian Naive Bayes missing = NaN samples per lettergreep = 7 N / O / Z N+O / Z N / Z N / O O / Z 0.388 0.612 0.552 0.515 0.633 Baseline 0.621 0.889 0.878 0.526 0.859 SVM: linear 0.424 0.625 0.598 0.489 0.637 SVM: rbf 0.608 0.913 0.930 0.590 0.912 AdaBoost 0.389 0.605 0.547 0.488 0.563 Gaussian Naive Bayes
CategoryPaQu CategoryUniversalDependencies CategorySpraakAccenten CategoryForcedAlignment