Paramaters and Hyperparameters

Question 1

What is a model parameter?

Answer 1

Variables internal to the neural network. Values can be estimated right from the data.

Values define the skill of the model and are not set manually

Question 2

Model parameters are required by the models to make predictions - true or false?

Answer 2

True

Question 3

Are model parameters saved as part of the learned model?

Answer 3

Yes

Question 4

What are some examples of model parameters?

Answer 4

Weights and Biases

Question 5

What are model hyperparameters

Answer 5

configurations external to the neural network - values cannot be estimated right from the data

Question 6

If you have to manually specify a parameter, it is a standard parameter or a hyperparameter?

Answer 6

hyperparameter

Question 7

When a deep learning algo is being tuned, what are you really tuning?

Answer 7

the hyperparameter - examples would include grid search or random search

Question 8

If there is no clear-cut way to find the best value in a hyperparameter, what are some of the approaches for doing so?

Answer 8

Rules of thumb, copy values used in other problems, or search for the best value by trial and error

Question 9

What is the difference between model parameters and model hyperparameters?

Answer 9

model parameters can be estimated from the data while hyperparameters cannot

Model hyperparameteres are often referred to as parameteres because they are the parts of the machine learning that must be manually set and tuned