Course Time and Place
Tu, Th 5:00 - 6:15 CSI 1121
getoor AT cs.umd.edu
office hours: Tue 2:30-4:00PM and by appointment
raaghav AT cs.umd.edu
office hours: Mon 1:30-3PM. Wed 2:30-4:00PM and by appointment.
Students should have a background in probability, logic and algorithms. Students should also have a strong programming background. An introductory AI class is encouraged, but will not be strictly enforced.
It appears there will be a fair number of students auditing the class. If you are auditing the class, please keep engaged (do the readings, participate in class discussions). Laptops are strongly discouraged during lecture. It is distracting both for the students around you, and for the lecturer. Given the late hour of the class, we will all need to be on our toes to keep things lively.
This will be adjusted a bit based on student interests and background, but the basic topics covered will be:
- Intro & Basics
- Version Spaces
- Parameter Estimation
- Supervised Learning/Classification
- Linear Classifiers
- Linear Regression
- Logistic Regression
- Naive Bayes
- Overfitting, Bias-Variance Trade-off, Discriminitive vs. Generative Models
- Non-linear Classifiers
- Decision Trees
- Neural Networks
- Nearest Neighbor
- Margin-based Approaches
- Computational Learning Theory
- Sample Complexity
- PAC learning
- VC Dimension
- Unsupervised Learning/Clustering
- Hierarchical Agglomerative Clustering
- Spectral Methods
- Density Estimation via Structured Models
- Graphical Models
- Bayesian Networks
- Markov Random Field
- Graphical Models
- Reinforcement Learning
- Other topics as time permits:
- Structured Inputs & Outputs
Unfortunately, there is no great choice of text available for this course. Mitchell is good, but a bit dated, other books cover some topics well, but leave others out entirely. All of these books will be made available on reserve in the CS Library.
by Tom M. Mitchell
Publisher: McGraw-Hill Science/Engineering/Math; (1997)
Pattern Classification (2nd Edition)
by Richard O. Duda, Peter E. Hart, David G. Stork
Publisher: Wiley-Interscience; 2nd edition (2000)
The Elements of Statistical Learning
by T. Hastie, R. Tibshirani, J. H. Friedman
Publisher: Springer Verlag; ( 2001)
Data Mining: Practical Machine Learning Tools and Techniques (2nd Edition)
by Ian Whiten and Eibe Frank
Publisher: Elsevier; (2005)
Please visit http://mailman.cs.umd.edu/mailman/listinfo/cmsc726 and subscribe to the class mailing list (cmsc726 AT mcfeely.cs.umd.edu).
There will be several homeworks and exercises. There will be a midterm and final exam. A major component of the class is an individually chosen, small group project.
|Midterm||Thursday, March 30, In-Class||25%|
|Final||Thursday, May 18, 4:00-6:00 pm||25%|
Students are encouraged to discuss the homework to understand the problems and reach a solution. However, each student must write down the solution independently. In other words, each student must understand the solution well enough in order to reconstruct it by him/herself. In addition, each student should write on the problem set the set of people with whom s/he collaborated.
Important note on the honor code: We occasionally reuse problem set questions from previous years' courses, we expect the students NOT to copy, refer to, or even look at the solutions in preparing their answers. It will be considered an honor code violation to intentionally refer to previous year's solutions (or other solutions available on the web) The purpose of problem sets in this class is to help you think about the material, not just give us the correct answers.
Unless otherwise stated, homeworks and projects are due in class on their due date. Due dates and times will be specified for each project. A grading penalty will be applied to late homeworks and projects. Recognizing that students may face unusual circumstances and require some flexibility in the course of the quarter, each student will have a total of seven free late (calendar) days to use as s/he sees fit. No additional individual extensions will be given. Once these late days are exhausted, any homework turned in late will be penalized at the rate of 25% per late day (or fraction thereof). Under no circumstances will a homework or project be accepted more than five days after its due date. Late days are from 5PM to 5PPM for homeworks. Late homeworks should be handed in to the TA. If not available (e.g., on weekends), write the date and time on the assignment and push it under Prof. Getoor's door. In all cases, for late homeworks, students should write down the time that the homework is turned in and the number of late days used. It is a honor code violation to write down the wrong time.
The majority of the grading will be done by the TA. If you think there has been a mistake in grading your homework or exam, please submit a regrade request explaining in writing, precisely and consicely, the grading error that has occurred, to the TA. Such request must be made no later than 1 week after the material in question was returned to the class. Any request to have an assignment regraded may result in the entire assignment in question being regraded, possibly resulting in a loss of points.
Any evidence of unacceptable use of computer accounts or unauthorized cooperation on tests, quizzes, or projects will be submitted to the Student Honor Council, which could result in an XF for the course, suspension, or expulsion from the University.
Students claiming an excused absence for an exam must apply in writing and furnish documentary support (such as from a health care professional who treated the student). Excused absenses do not extend your 7 late day budget.