def on_epoch_end(self, epoch, logs={}):
Callback.on_epoch_end(self, epoch, logs=logs)
# clear plot
self.axFig.cla()
# plot target embbeding
targetEmbeddingHandle = self.axFig.scatter(
self.targetEmbedding[:, 0],
self.targetEmbedding[:, 1],
marker="s",
alpha=0.25,
s=10,
c=self.yTarget,
cmap="rainbow",
label="MMD target embedding")
# plot network output projected on target embedding
plotPredictions = self.netAnchorLayerPredict(self.xInput)
# print(plotPredictions)
projection = np.dot(plotPredictions,
self.pca.components_[[0, 1]].transpose())
NetOuputHandle = self.axFig.scatter(
projection[:, 0],
projection[:, 1],
marker=">",
c=self.yInput,
cmap="rainbow",
alpha=0.25,
s=10,
label='Net output projected on target embedding')
self.axFig.legend(handles=(targetEmbeddingHandle, NetOuputHandle))
plt.draw()
plt.pause(0.01)
def on_epoch_end(self, epoch, logs={}):
Callback.on_epoch_end(self, epoch, logs=logs)
self.losses.append(logs.get('loss'))
self.val_losses.append(logs.get('val_loss'))
# plot the cost for training and testing so far
lossHandle, = self.ax.plot(self.losses, color='blue', label='loss')
val_lossHandle, = self.ax.plot(self.val_losses,
color='red',
label='validation loss')
self.ax.legend(handles=[lossHandle, val_lossHandle])
plt.draw()
plt.pause(0.01)
def train(self):
model = self.prepare_model()
training_info = self.get_training_info()
epoch_count = training_info['epoch']
def on_epoch_end(epoch, log):
nonlocal epoch_count
epoch_count += 1
self.store_training_info({
'epoch': epoch_count,
'val_acc': log['val_acc'],
'train_loss': log['loss'],
'val_loss': log['val_loss']
})
training_info_call_back = Callback()
training_info_call_back.on_epoch_end = on_epoch_end
model.fit(self.train_x, self.train_y,
batch_size=2, epochs=100,
validation_data=(self.test_x, self.test_y),
initial_epoch=epoch_count, callbacks=[ModelCheckpoint(self.model_weight), training_info_call_back])