Assignment #2: Critical Substantive Concepts of Machine Learning
Please complete the Module 2 readings before completing the assignment. Make sure that all responses are in your own words. Plagiarizing/copying and pasting from the Internet are against University policy.
1. In a 50+ word response, explain why Occam’s Razor is a vital principle in Machine Learning if your ML application is used for safety critical applications.
2. Explain briefly how bayesian methods are utilized in ML. (50 words)
3. Define and compare to each other: Ranking, Regression, Classification.
4. Explain why decision trees, one of the oldest methodologies in ML, are vital to sound ML learning algorithms.
5. What is Dimensionality Reduction and why is it used in ML?
6. What are Outliers and why is there detection and handling vital in ML algorithms and processes?
7. The text makes a vital declaration “Data starts to drive the operation; it is not the programmers anymore but the data itself that defines what to do next.” Why is this statement profound and how does it form the basis for using machine learning?
8. List and describe the Marr “Three Levels of Analysis”. How do these map to machine learning principles?
9. In machine learning, how do we model uncertainty? Why is this central in many machine learning algorithms?
10. Why is model selection vital in developing a ML solution? Give 2 examples of models used in ML.
11. Describe parametric estimation and non-parametric estimation. How are these used in ML?
12. Why in ML is our aim to fit a model to the data?
Respond in a Word document, all in your own words, and upload to Canvas.
Module 2 Readings
For the Module 2 readings, like in Module 1, we are focusing on the basic substantive concepts of Machine Learning as a subset discipline of Artificial Intelligence.
The Alpaydin MIT book is a great introduction to the topic, and over this and next Module we will be reading the book and highlighting the various components of machine learning. When reading the book, note the various components of ML: algorithms, data structures, mathematics, applications, etc. See how they are all related and are ultimately organized into a viable ML process, set of algorithms, and product in order to solve a specific problem.
Remember, in computer science our goal is to solve problems, whether they are mathematical, scientific, medical, business or related to another discipline.
Also remember Occam’s Razer when dealing with creating ML solutions. The simplist solutions are usually the best. (Especially when complexity increases software risk!!!). The brief overview is to remind you of this vital universal engineering (and life) principle.
The Brighterion white paper goes over the intersection between AI and ML. In this directed study we will read many white papers, as our goal with ML is to develop solutions to problems for industry. As a computer science and AI professional we must keep up with industry and its developments as well as the companies that are leading the AI/ML edge.
1. Read Alpaydin text: Pages 1-84
2. Read the Brighteron White paper on AI/ML
3. Read the brief explanation of Occam’s Razor.
