1
Q
Name the 3 parts we split the dataset into.
A
- Training
- Validation
- Test
2
Q
What is Overfitting?
A
When we have a model that scores well on training but poorly on validation and test datasets.
3
Q
What is Underfitting?
A
Where we have not captured the logic of the underlying data, leading to poor predictive power
4
Q
When is N-fold Cross Validation used?
A
When you have smaller amounts of data to be used to train on
5
Q
What is Early Stopping?
A
A technique to prevent overfitting
stop training early before we overfit
6
Q
How do we implement Early Stopping?
A
- Train for a preset number of epochs
- (better) stop when the loss function updates become too small
- (best) Validation set Strategy
7
Q
What is Validation Set Strategy?
A
