CSCI 386 - Private and Fair Data Analysis

This course studies two important social or ethical constraints one might face when analyzing data. The first half of the course covers privacy concerns, where one must analyze a data set without violating the privacy of the people whose data it contains. The second half of the course covers fairness, where one is seeking to ensure that classification rules output by a machine learning algorithm are not discriminatory with respect to race, gender, or other protected attributes. Throughout the course we will focus both on the process of creating mathematical formalizations of socially desirable properties, and on the design of algorithms that satisfy those definitions. Students will engage directly with current research papers.

Unit(s): 1
Group Distribution Requirement(s): Distribution Group III
Prerequisite(s): CSCI 121 or MATH 141; and one of CSCI 382/MATH 382, CSCI 387/MATH 387, or MATH 391
Instructional Method: Lecture
Grading Mode: Letter grading (A-F)
Cross-listing(s): MATH 386
Group Distribution Learning Outcome(s):

Use and evaluate quantitative data or modeling, or use logical/mathematical reasoning to evaluate, test or prove statements.
Given a problem or question, formulate a hypothesis or conjecture, design an experiment, and collect data or use mathematical reasoning to test or validate it.