Computer Simulation of an Artificial Neural Network
Using the Netlogo platform to run simulations of a basic neural network called the perceptron, students explore a basic, yet powerful, model of machine learning as they are challenged to understand the logic. Students engage in the perceptron model and discover a weakness of the model. The students then move on to run simulations on Netlogo with the multi-layer perceptron which overcomes the weakness in the original perceptron model.
Neural networks are modeled from the neural networks in a human brain.Copyright 2020 Chris Tyler, Michigan State University RET
Machine learning is the process by which a computer is able to improve its own performance by continuously incorporating new data into an existing statistical model.
Computer scientists use machine learning to automate the process of data analysis in a variety of industries. In the financial sector, engineers apply principles of machine learning to engineer systems that can gain insights into complex data sets to prevent fraud. Healthcare professionals can use machine learning to help improve diagnoses and treatments. Computer scientists must understand the logic of the particular machine learning model and mathematics involved in machine learning in order to choose the appropriate model and parameters.
In this activity, students play the role of the computer scientist by comparing the logic of two machine learning models including their strengths and weaknesses.
After this activity, students should be able to:
- Identify the parameters used in the perceptron model.
- Describe the process the perceptron uses to learn a rule.
- Identify the major weakness of the single perceptron model.