Topic 11e
Artificial Neural Networks
Learning Outcomes
By the end of this topic students should be able to:
- Explain the concept of a perceptron.
- Given a simple perceptron model and set of inputs, identify the output of the perceptron.
- Given a simple perceptron model, explain the function of the perceptron (explain its outputs).
- Construct a simple perceptron model to classify a provided dataset.
- Explain how multiple perceptrons are combined to form an artificial neural network (ANN).
Learning Materials
- Readings
- Artificial Neural Networks - pp 593-598
- Videos
- Crash Course AI #2 - Neural Networks
- Today you should finish this video
- Watch from 2:46-end
- Today you should finish this video
- Crash Course AI #3 - Neural Networks and Deep Learning
- Crash Course AI #6 - Unsupervised Learning
- In some ways this general topic could have happened as part of Topic 11b. But, this video frames it in the context of Neural Networks. So we put it here instead.
- If you feel like you would benefit from hearing/seeing some of this again, but in a different way, we highly suggest these videos:
- Neural Network Examples (From the Crash Course in Statistics)
- Neural Networks Explained (Goes in to the math/structure of the network in more detail)
- Activities
- Perceptron Worksheet
- Single Perceptron Visualizer
- While this is very mathematical, you can focus on the corners of the visualizer and play with how the weights on the two inputs (wx and wy) interact with the bias/threshold (b) to change the outputs.
- Tensor Flow Playground
- Observe how different neural networks can be trained on different data configurations.
Checking for Understanding
Answer the following questions from your textbook. The answers to all Q&E questions are in the back of your book in Appendix F.
- p 597, #1
- p 598, #2
Additional Guidance
- NA
Further Information
- Crash Course AI #4 - Training Neural Networks
- This video explains how neural networks actually get trained. It's a bit mathy (more so than we feel you need for the learning outcomes in this course). But it does a great job of explaining this material at a level that even your high school students could understand.
- Crash Course AI #5 - You can train a Neural Network to read your handwriting
- This video introduces a Colab Program that you can use to actually train a neural network. If you don't know anything about programming, this is way too much. But if you know a little bit you (or your students) might be able to do some pretty cool things with this.