def vari_confidence_ER(z, cutoff):
x, y = vari_confidence_mult(z, cutoff, 0)
x = np.array(x)
b, m = polyfit(x, y, 1)
plt.plot(x, y, 'ro', markersize=3)
save_plot_custom("confidence ER")
plt.clf()
def loss(history):
history_dict = history.history
loss_values = history_dict['loss']
val_loss_values = history_dict['val_loss']
epochs = range(1, len(loss_values) + 1)
plt.plot(epochs, loss_values, 'bo', label='Training loss')
plt.plot(epochs, val_loss_values, 'b', label='Validation loss')
plt.title('Training and validation loss')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
save_plot_custom("Loss")
def accuracy(history):
history_dict = history.history
acc_values = history_dict['acc']
val_acc_values = history_dict['val_acc']
plt.clf()
loss_values = history_dict['loss']
epochs = range(1, len(loss_values) + 1)
plt.plot(epochs, acc_values, 'bo', label='Training acc')
plt.plot(epochs, val_acc_values, 'b', label='Validation acc')
plt.title('Training and validation Accuracy')
plt.xlabel('Epochs')
plt.ylabel('acc')
plt.legend()
save_plot_custom("Accuracy")
def colored_contamination_ER(sizes, predictions, labels, cutoff):
predictions = predictions.reshape(predictions.shape[0], )
dset = list(zip(sizes, predictions, labels))
bins, conf = vari_confidence_mult(dset, cutoff, ER_LABEL)
contamination = []
acontamination = []
for i in range(len(bins)):
contamination = contamination + list(
filter(
lambda z: z[0] == bins[i] and z[1] < conf[i] and z[2] ==
NR_LABEL, dset))
acontamination = acontamination + list(
filter(
lambda z: z[0] == bins[i] and z[1] > conf[i] and z[2] ==
NR_LABEL, dset))
if contamination:
x = list(zip(*contamination))[0]
y1 = list(zip(*contamination))[1]
else:
x, y1 = [], []
if acontamination:
ax = list(zip(*acontamination))[0]
ay1 = list(zip(*acontamination))[1]
else:
ax, ay1 = [], []
plt.plot(x,
y1,
'go',
markersize=3,
label='NR in ' + str(cutoff) + ' ER confidence')
plt.plot(ax,
ay1,
'bo',
markersize=3,
label='NR outside ' + str(cutoff) + ' ER confidence')
plt.plot(bins,
conf,
'ro',
markersize=3,
label=str(cutoff) + ' ER confidence')
plt.legend(loc='best')
save_plot_custom("colored_contamination_ER")
plt.clf()
def NR_contamination(sizes, predictions, labels, cutoff):
predictions = predictions.reshape(predictions.shape[0], )
dset = list(zip(sizes, predictions, labels))
bins, conf = vari_confidence_mult(dset, 1 - cutoff, NR_LABEL)
contamination = []
for i in range(len(bins)):
contamination = contamination + [
len(
list(
filter(
lambda z: z[0] == bins[i] and z[1] > conf[i] and z[2]
== ER_LABEL, dset)))
]
plt.plot(bins,
contamination,
'go',
markersize=3,
label='Number of contamination points')
plt.legend(loc='best')
save_plot_custom("contamination_NR " + str(sum(contamination)) +
" contamination points")
plt.clf()