Bell’s workshop.
How can a neural network improve its predictions?
In the previous lesson, we trained a neural network to identify round tables using only their area and radius.
The network learned from examples and was able to make real predictions.
But what if we are not fully convinced by its prediction capability?
Learning from an experiment
The Bell’s Workshop scenario is an experiment mainly focused on showing how to improve the prediction from the first lesson.
By experimenting with the scenario, we will understand what the following is:
More neurons
How can the prediction improve if we add more hidden neurons to the neural network?
Neural network architecture
See for the first time how the neural network grows.
Accuracy
See how increasing the number of neurons improves prediction accuracy.
How do you experiment with the scenarios?
It all begins when you find yourself in a scenario with a problem that you need to solve.
In this case, we find ourselves in the Bell’s Workshop scenario, and the problem we need to solve is determining whether a table is round or not with greater accuracy by increasing the size of the neural network.
Progress through the sections and see for yourself, visually, how we can improve the training and ultimately achieve near-perfect accuracy.
01
Reflection
02
Discovery
03
Entanglement
Continue the experiment