- Instructor: Prof.
Sung-Hyuk Cha
- Email: scha@pace.edu
- Tel: (212) 346-1253
- Office: 163 William St. 2nd floor rm 228
-
Office Hours: 4:00pm~5:40pm on Monday, Tuesday, and Thursday
- CRN: 23764
- Meeting:
- Meeting Times: T 06:00 - 09:15 PM,
Spring 2021
- Place: CIVIC E304
- Textbook:
Understanding Machine Learning: From Theory to Algorithms by
Shai Shalev-Shwartz and Shai Ben-David,
Cambridge University Press, 2014, ISBN 978-1107057135
- Description:
This course teaches students machine learning theory and algorithms. Students will learn about probably approximately correct (PAC), empirical risk minimization (ERM), structural risk minimization (SRM), and minimum description length (MDL) learning rules. Students will then study various machine learning algorithms, such as linear models, gradient descent, support vector machines (SVM), kernel methods, and trees, and how they connect to the theoretical framework. Finally the course culminates with additional topics such as clustering, dimensionality reduction, generative models, and feature selection.
- Learning Outcomes: After taking this course students will:
- Understand the theoretical framework for various learning rules
- Understand the various algorithms used in machine learning
- Connect the theoretical issues to algorithms and know how to outline algorithms for learning problems
- Prerequisites:
CS 660 with a minimum grade of C
- Lecture Notes: can be accessed using the http://classes.pace.edu
- Tentative Schedule: to be fixed.
| Week |
Topic |
| 1 (1/26) |
Ch 18 Decision trees
|
| 2 (2/2) |
Ch 19 Nearest Neighbor
|
| 3 (2/9) |
Ch 9 Linear predictors
|
| 4 (2/16) |
Ch 10 Boosting
|
| 5 (2/23) |
Ch 22 Clustering
|
| 6 (3/2) |
Ch 23 Dimensionality reduction
|
| 7 (3/9) |
Midterm |
| 8 (3/16) |
Ch 3 A formal learning model &
Ch 4 Learning via uniform convergence
|
| 9 (3/23) |
Ch 5 The bias-complexity trade-off
Ch 6 The VC-dimension
Ch 7 Non-uniform learnability
|
| 10 (3/30) |
Ch 11 Model selection and validation
Ch 12 Convex learning problems
Ch 13 Regularization and stability
|
| 11 (4/6) |
Ch 14 Stochastic gradient descent
|
| 12 (4/13) |
Ch 15 Support vector machines
Ch 16 Kernel methods
|
| 13 (4/20) |
Ch 24 Generative models
Ch 25 Feature selection and generation |
| 14 (4/27) |
Final |
- Evaluation:
- Assignments (50%):
- Midterm Exam (20%):
- Final Exam (30%):
- Accommodations for Students with Disabilities
The University's commitment to equal educational opportunities for students with disabilities includes providing reasonable accommodations for the needs of students with disabilities. To request a reasonable accommodation for a qualified disability a student with a disability must self-identify and register with the Office of Disability Services for his or her campus. No one, including faculty, is authorized to evaluate the need for or grant a request for an accommodation except the Office of Disability Services. Moreover, no one, including faculty, is authorized to contact the Office of Disability Services on behalf of a student. For further information, please see Resources for Students with Disabilities at www.pace.edu/counseling/resources-and-support-services-for-students-with-disabilities.
- Academic Integrity:
(From the Student Handbook)
Students are required to be honest and ethical in satisfying their academic assignments and requirements.
Academic integrity requires that, except as may be authorized by the instructor, a student must demonstrate independent intellectual and academic achievements.
Therefore, when a student uses or relies upon an idea or material obtained from another source, proper credit or attribution must be given.
A failure to give credit or attribution to ideas or material obtained from an outside source is plagiarism. Plagiarism is strictly forbidden.
Every student is responsible for giving the proper credit or attribution for any quotation, idea, data, or other material obtained from another source that is presented (whether orally or in writing)
in the student's papers, reports, submissions, examinations, presentations and the like. Individual schools and programs may have adopted additional standards of academic integrity. Therefore, students are responsible for familiarizing themselves with the academic integrity policies of the University as well as of the individual schools and programs in which they are enrolled. A student who fails to comply with the standards of academic integrity is subject to disciplinary actions such as, but not limited to, a reduction in the grade for the assignment or the course, a failing grade in the assignment or the course, suspension and/or dismissal from the University.
|