Neural Network Learning Flashcards
1
Q
What is the blame assignment problem?
A
Calculating the measure of how much each neuron contributed to the overall error of the network at the output layer.
2
Q
What is the difference between an epoch and an iteration?
A
An epoch is a single pass through all the examples in the training dataset.
An iteration is a single forward and backward pass plus weight update of the backpropagation algorithm.
3
Q
Complete: Updating weights after each training example is known as ___, _____, or ____ gradient descent.
A
on-line
sequential
stochastic
4
Q
Explain mini-batch gradient descent.
A
Dataset is split into multiple subsets called mini-batches.
