train: each iteration over all the training data is called an epoch
layers:
dense: vector data (samples, features)
recurrent: sequence data (samples, timesteps, features)
image: 2D convolution
layer input and output: 784 input, 32 output
from keras import models
from keras import layers
model = models.Sequential()
model.add(layers.Dense(32, input_shape=(784,)))
model.add(layers.Dense(32))# input 32
topology of a network defines a hypothesis space
network topology define the tensor operation series
for multi-loss networks, all losses are combined via a function (average) into a single scalar quantity, for serving gradient-descent process
loss function:
binary crossentropy for two-class classification
categorical crossentropy for many class classification
mean squared error for regression problem
connectionist temporal classification for sequence-learning problem