| Contact Information | Course Description | Course Goals and Objectives | Course Outline | Calendar |
| Course Materials | Grading | Attendance and Participation | Assignments | Honesty |
Contact Information
| Professor | Richard J. Povinelli, Ph.D. |
| richard.povinelli@marquette.edu | |
| Homepage | http://povinelli.eece.mu.edu |
| D2L | https://d2l.mu.edu/ |
| Phone | 414.288.7088 |
| Office Hours | Haggerty Hall 221 |
| Extra | Can be deleted |
Course Description and Prerequisites
Machine Learning is the study of computer algorithms that improve automatically through experience (Mitchell 1997). In this class we will study both the algorithms themselves and the theoretical foundations of this field. Machine learning algorithms to be studied including decision trees, artificial neural networks, Bayesian learners, evolutionary algorithm, and boosting and bagging techniques. We will study a computational learning theory and PAC learnablity.
Course Goals
- By the end of this course, you should...
- Be able to explain and apply the constituent technologies that form machine learning.
Course Objectives
- By the end of this course, you should...
- Be able to formulate and pose well-structured machine learning problems.
- Be able to explain various searching approaches and analyze their tradeoffs.
- Be able to explain what a hypothesis space is.
- Be able to program a decision tree learning algorithm.
- Be able to explain the capabilities of ID3 and C4.5 and how they use pruning.
- Be able to explain and use artificial neural networks, including the back propagation algorithm and other alternative methods for training a neural network.
- Be able to determine the information gain provided by an attribute in a decision tree.
- Be able to summarize and apply computational learning theory.
- Be able to use various Bayesian learning approaches including Naïve Bayes methods.
- Be able to explain the assumptions made by the Naïve Bayes methods.
- Be able to build a simple evolutionary algorithm and explain its operators.
- Be able to create a resolution rule learner.
- Be able to explain how a support vector machine works.
- Analytic learning, both prior knowledge and training examples
- Be able to explain the theory behind ensemble learners.
- Be able to create and apply boosting and bagging algorithms.
- Be able to analyze the computational complexity of a machine learning algorithm.
Course Outline
| What | When |
| Machine Learning Framework | wk1-2 |
| Decision Trees | wk3-4 |
| Artificial Neural Networks | wk5-6 |
| Computational Learning Theory | wk7-8 |
| Support Vector Machines | wk9-10 |
| Combining Multiple Learners | wk11-12 |
| Dimensionality Reduction | wk13 |
| Reinforcement Learning | wk14 |
| Presentations | wk15 |
Course Materials
Required Texts
- Foundations Of Machine Learning second edition by Mehryar Mohri, Afshin Rostamizadeh and Ameet Talwalkar, MIT Press, 2018.
Recommended Texts
- Machine Learning: An Algorithmic Perspective by Stephen Marsland, Chapman and Hall/CRC, 2009.
- Introduction to Machine Learning by Ethem Alpaydin, MIT Press, 2004.
- Machine Learning by Tom Mitchell, McGraw Hill, 1997.
- The Elements of Statistical Learning: Data Mining, Inference, and Prediction by Trevor Hastie, Robert Tibshirani, Jerome Friedman, Springer, 2001.
- Learning Kernel Classifiers by Ralf Herbrich, MIT Press, 2001.
Grading
| What | Number | Value per | Total |
| Article Reviews | 5 | 1.8% | 9% |
| Peer Review of Article Reviews | 5 | 0.2% | 1% |
| Homework | 5 | 3% | 15% |
|
1 | 1% | 1% |
| Project Paper Draft | 2 | 2% | 4% |
| Project Paper Final | 1 | 20% | 20% |
| Paper Presentation Final | 1 | 5% | 5% |
| Midterm Exam | 1 | 20% | 20% |
| Final Exam | 1 | 25% | 25% |
| Total | 100% |
NOTE: All dates and numbers are subject to change as deemed necessary!
Grade Scale
| 93+ | A |
| 90-93 | A- |
| 87-90 | B+ |
| 83-87 | B |
| 80-83 | B- |
| 77-80 | C+ |
| 73-77 | C |
| 70-73 | C- |
| 67-70 | D+ |
| 60-67 | D |
| 0-60 | F |
The grading scale is the most stringent one you will be held to, i.e. I can give you a higher letter grade than shown on the scale, but never a lower one. If you have missing assignments, you are inelligible to receive a higher grade.
Late Assignments
I will deduct 5% for assignments up to one day late, 10% for two days late, and 15% for up to three days late, and so on up to a maximum of 50% off. Assignments are due at the beginning of class. They are late after that. Assignments are not accepted after solutions have been distributed, nor after the last day of class. In class assignments are only accepted during the class period they are assigned.
Attendance and Participation
I have always enjoyed teaching classes where the students actively participate - a conversation is more fun than a monologue! Although there is no specific credit assigned for attending, it is still expected. There may be in class graded assignments. These may be turned in only during the class period they are given.Assignments
You should expect to spend, on average, from twelve (12) to fifteen (15) hours on article reviews, critiques, reading, homeworks, and other preparation for this class. This time is in addition to the three (3) hours of lecture you are expected to attend every week.
All written portions of assignments must be created using a word processor. No part of the writeup may be hand drawn. The assignments are to be well written with proper spelling and grammar. Points will be deducted for poorly written assignments. Written portions of assignments must be turned in as MS Word documents (.docx format). Code and other portions must be submitted in the proper electronic format. I will deducted 5% from incorrectly formatted assignments.
All assignments must be turned in via D2L. Assignments are due according to the the time specified in the D2L calendar.
Article Reviews
There will be five (5) article reviews. This will help you understand the relevant literature for your research project. You will identify articles that are relevant to the course topic, if you are unsure of its relevance double check with the instructor. For each article you will write a 1 page summary and a 1 page critique.
Peer Review of Article Reviews
Critiquing others work is an excellent mechanism for improving your own. It is backbone of the peer review process. You will be expected to evaluating others work including critiquing others article reviews.
Homeworks
There will be five (5) homeworks, which will be collected and graded. The homework assignments will be scaled to 400 points. The homeworks will be combinations of questions from the book and programming problems. You must cite, using IEEE format , your references including online ones.
Research project
There will be one (1) research projects. This will be an original research effort with the potential of resulting in a conference quality paper and associated presentation. You will be expected to propose, execute, write up, and present independent and original research.
Exams
There will be a midterm exam worth 20% of your grade. There will be a final exam worth 25% of your grade.