Day 22 Artificial Neural Networks 2¶

Dec 3, 2020¶

CMSE logo

Announcements¶

  • Homework 5 Working with Tensorflow. Due this Friday. This is the last homework assignment!
    • Having trouble installing tensorflow? Use Google Colab (just upload your notebook).
  • Projects Rubric posted to D2L.
    • Due Dec 14th; Review 3 projects by Dec 16th
    • 8-10 minute video presentation + documented notebook on your analysis
    • 3 In-class work periods for the project

Calendar¶

This week¶

  • Thursday 12/3: Day 21 Neural Networks 2

Last week of classes¶

  • Tuesday 12/8: Project work day 3
  • Thursday 12/10: Project work day 4

Artificial Neural Networks¶

Today¶

  • We are going to work with a code base that implements a Neural Network class.
  • You will make some changes to the code base to make it more general.
  • You and your groups will need to read through the code base to understand how its works.

Let's go over the pre-class¶

Assume you have a set of feature (features) and labels (labels).

Using this code base, we will create a Neural Network instance.

NN = Neural_Network()

We can then perform forward propagation using the .forward method.

NN.forward(features)

Note: For the existing code base, the Neural Network class has specified values for input layers, hidden layers, and output layers.

How do we train a model?¶

Again, assume you have a set of feature (features) and labels (labels).

Using this code base, we will create a trainer instance.

T = trainer(NN)

We can then train the model on our data using the .train method.

T.train(features, labels)

Note: You need to pass the Neural Network instance to create a trainer instance.

Questions, Comments, Concerns?¶