here we use activation as sigmoid … so what is relu??? have i missed anything?? and for loss we use epoch so here what is ‘binary_crossentropy’. ? and in previous videos we saw model as neural network here we use model as sequential???what is sequential model?
N this video what is relu?
Relu is also an activation function which is highly preferred over other functions these days. It is defined as y = max{0,x}.
Advantages of relu:
- Biological plausibility: One-sided, compared to the antisymmetry of tanh.
- Sparse activation: For example, in a randomly initialized network, only about 50% of hidden units are activated (having a non-zero output).
- Better gradient propagation: Fewer vanishing gradient problems compared to sigmoidal activation functions that saturate in both directions.
- Efficient computation: Only comparison, addition and multiplication.
.
Epoch is a totally different concept as compared to a loss function. Epochs refer to the number of iterations over the entire training dataset which we undergo whole training a model. While, binary_crossentropy is a loss function which we use in the case of two y classes only.
.
The simplest model in an Artificial Neural Network is defined in the Sequential class which is a linear stack of Layers.