Description
The projects can be on almost any topic in Machine Learning.
- It can be a topic that we covered (or will be covering) in class
- It can be a topic that we didn’t cover (or won’t be covering) in class.
- You can cover a method in machine learning.
- You can apply a method (or methods) from machine learning to a dataset that you find interesting.
The project must be mathematical and/or quantitative in nature. For example, if your subject is the history of a machine learning method to solve a particular problem, you’ll need to go through the equations and perhaps even develop and run some code. In short, it need to be a computer science project and not (just) a history of computer science project.
If you borrow ideas or code from anyone or any online (or offline) source, make sure to credit that source.
1
If you choose a topic that we covered in class you will need to go beyond what we went over in class.
Here are some things you may want to include in your project. Some will be relevant to some topics, but not to others:
- How does how long your method take as you change the size of the input; that is, what is it’s O? Compare the theoretical O with what you observe.
- How accurate is your method? How does the accuracy vary as you change network design or key hyperparameters?
- How do the timings and/or accuracy vary across different methods to solve the same problem?
