Perceptron algorithm
Today, you will be introduced to the perceptron algorithm, which is commonly used for supervised learning of binary classifiers. The perceptron is an important algorithm because it constitutes the basis of what we nowadays call deep learning - i.e. it is a single-layer neural network.
0) Learning Objectives:
- Identify the main elements of Rosenblatt's perceptron
- Determine what kind of problems can, and cannot be solved with Rosenblatt's perceptron
- Develop a code implementation of Rosenblatt's perceptron
Table of contents:
- Introduction: 1 hour
- Workshop: 4 hours
- Additional material (optional): 2 hours 3.1 The ‘original' perceptron by Rosenblatt
Questions or issues?
If you have any questions or issues regarding the course material, please first ask your peers or ask us in the Q&A in Datalab!
Good luck!
1) Introduction
Some people think of Perceptor every time they hear the word perceptron. Both are linked to the field of science (i.e. The Perceptor is a scientist, while the perceptron was developed by a scientist, Frank Rosenblatt). However, they are quite distinct. For example, I think the perceptron is way cooler than the Perceptor. What do you think?
Figure 1. The perceptron vs. The perceptor.
1a Watch the video Supervised learning: The perceptron by Crash Course.
Video 1. Supervised learning: The perceptron by Crash Course.
1b Define the term ‘perceptron'. Write your answer down (maximum of 100 words).
1c Provide at least one advantage, and one limitation of the perceptron algorithm. Elaborate on your answer.
2) Workshop
Now, we're introduced to the perceptron it's time to ground down these fundamentals by doing a workshop. Open the Basics of Machine Learning course on Codecademy and complete the module: Perceptron, specifically:
- Lesson: Perceptron
- Quiz: Perceptron
- Project: Perceptron Logic Gates
3) Additional material (optional)
3.1 The ‘original' perceptron by Rosenblatt
This part of the independent study material is optional. You are not required to complete the following questions. However, they may help you with understanding how the perceptron algorithm works.
3a Read The Perceptron - A Guided Tutorial Through Its History and Implementation In Python by Pablo Caceres.
3b The perceptron can be decomposed into three main elements. List, and subsequently describe them. Write your answer down (maximum of 100 words).
3c Write down the mathematical equation that represents the decision-boundary or ‘hyperplane'.
3d After completing the Codecademy workshop, and the perceptron tutorial by Pablo Caceres, would you provide the same answer to exercise 1b? Explain your answer.
Next up!
Coming Datalab we will reflect on the perceptron again and give you an opportunity to ask any questions you might have.
In Datalab, we will apply the perceptron algorithm to the Oosterhout dataset in the context of our problem statement!