Stage $4$

Possible Solution

Creating an ANN model depends heavily on the number of layers in the network. To construct the network, you must determine a number of parameters, including the number of layers, the number of neurons in each layer, the activation function required, and the number of outputs.

Remember that the number of layers in an ANN, and the number of neurons inside each layer you can decide it based on the task type.

As you have learned in the student book, follow the steps below to design an ANN model.

STAGE $5$ Final solution $/$ Python coding

Activity $2\mathrm{.}5\mathrm{.}1$

After you have suggested and tried different ANN network structures, you need to choose the code that showed the best accuracy to complete your project.

Answering the following questions might help you with your conclusion:

Which of the ANN models had the simplest structure?

Which activation function was used for each model?

What was the prediction accuracy of each model?

How long did each model take to make a prediction?

Determine which scenarios $($from Stage $4)$ satisfy the design brief, and which one is the best.

State at least one advantage and one disadvantage for each scenario you tested.