words = draw_words(stimuli, bars)
stacked = np.concatenate([words1, words2, words], axis=1)
# pad this slightly in order to be able to distinguish groups
stacked = pad(stacked, [0, 10, 10])
num_x = 8
num_y = 12
start_at = 0
collage = make_collage(stacked[start_at:start_at + (num_x * num_y)].\
reshape(num_x, num_y, stacked.shape[1], stacked.shape[2]))
import pylab as pl
pl.figure()
pl.imshow(collage)
pl.gray()
import scoring
pl.figure()
roc1 = scoring.roc(res1, stimuli)
roc2 = scoring.roc(res2, stimuli)
pl.plot(roc1)
pl.plot(roc2)
pl.plot([0, len(stimuli)], [0, 1])
pl.show()
0.5 * ((y_true == y_pred) * (1 - y_true)).sum() / (1 - y_true).sum()
from sklearn.cross_validation import cross_val_score
scores1 = cross_val_score(forest,
res1 > .5,
stimuli,
cv=6,
score_func=score_func)
#scores2 = cross_val_score(, res1 > .5, stimuli, cv=6,
# score_func=score_func)
import pylab as pl
pl.figure()
pl.imshow(collage)
pl.gray()
import scoring
pl.figure()
roc1 = scoring.roc(p, stimuli)
roc2 = scoring.roc(predictions, stimuli)
pl.plot(roc1, c='b', label='Mot entier')
pl.plot(roc2, c='g', label='Lettre par lettre')
pl.grid()
pl.title(
'Résultats de deuxième couche, avec une première couche logistique L2 et k=100 voxels'
)
pl.plot([0, len(stimuli)], [0, 1])
pl.show()
words2 = draw_words(res1, bars)
words = draw_words(stimuli, bars)
stacked = np.concatenate([words1, words2, words], axis=1)
# pad this slightly in order to be able to distinguish groups
stacked = pad(stacked, [0, 10, 10])
num_x = 8
num_y = 12
start_at = 0
collage = make_collage(stacked[start_at:start_at + (num_x * num_y)].\
reshape(num_x, num_y, stacked.shape[1], stacked.shape[2]))
import pylab as pl
pl.figure()
pl.imshow(collage)
pl.gray()
pl.show()
import scoring
pl.figure()
roc1 = scoring.roc(res1, stimuli)
#roc2 = scoring.roc(res2, stimuli)
pl.plot(roc1)
#pl.plot(roc2)
pl.plot([0, len(stimuli)], [0, 1])
words = draw_words(stimuli, bars)
stacked = np.concatenate([words1, words2, words], axis=1)
# pad this slightly in order to be able to distinguish groups
stacked = pad(stacked, [0, 10, 10])
num_x = 8
num_y = 12
start_at = 0
collage = make_collage(stacked[start_at:start_at + (num_x * num_y)].\
reshape(num_x, num_y, stacked.shape[1], stacked.shape[2]))
import pylab as pl
pl.figure()
pl.imshow(collage)
pl.gray()
pl.show()
import scoring
pl.figure()
roc1 = scoring.roc(res1, stimuli)
pl.plot(roc1)
pl.plot([0, len(stimuli)], [0, 1])
stacked = np.concatenate([words1, words2, words], axis=1)
# pad this slightly in order to be able to distinguish groups
stacked = pad(stacked, [0, 10, 10])
num_x = 8
num_y = 12
start_at = 0
collage = make_collage(
stacked[start_at : start_at + (num_x * num_y)].reshape(num_x, num_y, stacked.shape[1], stacked.shape[2])
)
import pylab as pl
pl.figure()
pl.imshow(collage)
pl.gray()
pl.show()
import scoring
pl.figure()
roc1 = scoring.roc(res1, stimuli)
roc2 = scoring.roc(res2, stimuli)
pl.plot(roc1)
pl.plot(roc2)
pl.plot([0, len(stimuli)], [0, 1])
reshape(num_x, num_y, stacked.shape[1], stacked.shape[2]))
def score_func(y_true, y_pred):
return 0.5 * ((y_true == y_pred) * y_true).sum() / y_true.sum() +\
0.5 * ((y_true == y_pred) * (1 - y_true)).sum() / (1 - y_true).sum()
from sklearn.cross_validation import cross_val_score
scores1 = cross_val_score(forest, res1 > .5, stimuli, cv=6,
score_func=score_func)
#scores2 = cross_val_score(, res1 > .5, stimuli, cv=6,
# score_func=score_func)
import pylab as pl
pl.figure()
pl.imshow(collage)
pl.gray()
import scoring
pl.figure()
roc1 = scoring.roc(p, stimuli)
roc2 = scoring.roc(predictions, stimuli)
pl.plot(roc1, c='b', label = 'Mot entier')
pl.plot(roc2, c='g', label = 'Lettre par lettre')
pl.grid()
pl.title('Résultats de deuxième couche, avec une première couche logistique L2 et k=100 voxels')
pl.plot([0, len(stimuli)], [0, 1])
pl.show()
