Batch gradient descent - AshokBhat/ml GitHub Wiki

Batch gradient descent

What is batch gradient descent?
- Gradient descent where all input training examples are used in each iteration
How is it different from gradient descent?
- It is a form of gradient descent, where all input training examples are used in each iteration
How is it different from Stochastic gradient descent?
- Stochastic uses only 1 random example, whereas batch uses all
How is it different from Mini-batch gradient descent?
- Mini-batch uses a subset of input examples, not all.
When is it used?
- Small enough input datasets
When is it not used?
- Huge input datasets

How does it affect the path to minima?
- With batch gradient descent, the path to the minima is usually the shortest.