| Course ID: | CSCI 8960. 4 hours. |
| Course Title: | Privacy-Preserving Data Analysis |
Course
Description: | An introduction to the privacy preservation problems, as well as
algorithmic and statistical techniques for data privacy, in
modern data analysis, such as machine learning and data mining.
Approaches include randomized algorithms, synthetic data
generation, stability analysis, and so on. |
| Oasis Title: | Privacy-Preserving Data Analys |
| Prerequisite: | CSCI 4380/6380 or permission of department |
Semester Course
Offered: | Not offered on a regular basis. |
| Grading System: | A-F (Traditional) |
|
| Course Objectives: | The goal of this course is to introduce students to the privacy
preservation problems in modern data analysis, such as machine
learning and data mining. The course will explore mathematically
rigorous algorithmic and statistical tools that enable analysis
of sensitive data while protecting privacy of individuals. The
course is appropriate for students preparing to do research in
machine learning and data mining, as well as for Science and
Engineering students who want to learn how to deal with privacy
issues related to their research. |
| Topical Outline: | Review of linear algebra and probability theory
Privacy notions:
k-anonymity
l-diversity
Differential privacy (overview)
Differential privacy:
-laplace mechanisms
-exponential mechanism
Randomized algorithms
Linear query answering mechanisms:
-interactive VS non]interactive mechanisms
Synthetic database generation:
-histogram based approaches
-Bayesian network based approaches
Algorithmic stability analysis
Private convex optimization:
-objective perturbation
-posterior sampling |