Artificial Intelligence Methods for Social Good (Spring 2018)

Poster [jpg]

Basic Information

Course Number: 08-537 (9-unit) and 08-737 (12-unit), Spring 2018

Time: Tuesday/Thursday 10:30am-11:50am

Location: GHC 5222

Instructor: Fei Fang

Course Description

The rapid advance in artificial intelligence (AI) has opened up new possibilities of using AI to tackle the most challenging societal problems today. This course brings together a set of advanced AI methods that allow us to address such challenges and promote social good:

1)    Machine Learning: supervised learning, deep learning
2)    Game Theory and Mechanism Design: security games, human behavior modeling, scheduling and pricing, citizen science
3)    Sequential Decision Making: Markov Decision Processes (MDPs), partially observable MDPs
4)    Planning and Optimization: influence maximization, online planning, combinatorial optimization

In addition to providing a deep understanding of these methods, the course will introduce which societal challenges they can tackle and how, in the areas of (i) healthcare, (ii) social welfare, (iii) security and privacy, (iv) environmental sustainability. The course will also cover special topics such as AI and Ethics and Safety of AI. Example research projects and social good outcomes can be found at http://aiandsocialgood.org

The course content is designed to not have too much overlap with other AI courses offered at CMU. Although the course is listed within SCS, it should be of interest to students in several other departments, including ECE, EPP and SDS.

(9 Unit) The students in this 9-unit course are expected to have taken at least two mathematics courses covering linear algebra and probability. The students will work in groups on a systematic literature review or a project exploring the possibility of applying existing AI tools to a societal problem, with a survey paper or technical report and presentation delivered at the end of the semester.

(12 Unit) This 12-unit course is only open to graduate students (master and Ph.D. students) with previous programming experience and background knowledge in artificial intelligence. The students will work in groups on a research project with a research-style paper and an oral presentation delivered at the end of the semester. Please see the instructor if you are unsure whether your background is suitable for the course.

Learning Objective

At the end of the course, the students should be able to

  • Identify societal challenges that can potentially be tackled by AI methods, and determine which AI methods can be applied
  • Describe the AI methods covered in the course, including the basic concepts, the key algorithms, and the commonly-used implementation of the methods
  • Model the societal challenges as mathematical problems that AI techniques can be applied and propose how to adjust and modify the AI techniques to fit the problems
  • Describe evaluation criteria and methodologies of applying AI methods for social good
  • Deliver written and oral presentation on research projects or research survey

Class Format

This course will consist of traditional lectures given by the instructor, guest lectures given by key researchers who have extensive experience in “AI for Social Good”. The focus of each lecture will be the AI methods, or the application of AI for social good, or both. There will be a class discussion session every week based on the paper reading assignment.

Grading and Workload

The course will be letter graded. The amount of workload is indicated by the unit of the course. See a detailed breakdown below. The grade will depend on the following:

  • Class participation. Students are expected to be in class on time and attend at least 90% of the lectures. The grading of the class participation will be mostly based on attendance checked through in-class quizzes that will not be scored. Other factors include asking and answering questions in class.
  • Paper reading assignment. The course will require all students to do weekly paper reading assignments, and the students will submit a summary including an overview of the paper and a list of questions (due midnight before the day of the class). There will be 14 paper reading assignment in total. The lowest score will be dropped.
  • Biweekly written answers assignment. There will be 8 written answers assignment in total. The written answers assignment will involve checking the understanding of basic concepts, working through the algorithms presented in class on example problems and proposing potential applications of the AI methods covered in class or potential AI-based tools for the societal challenges covered in class. The lowest score will be dropped.
  • Final project. The students will work in small groups (2-3 students in each group). The instructor will provide some ideas about the project, and the students can also propose their own projects or literature review topics related to AI and Social Good but will need consent from the instructor. Project proposals (200 words) are due on February 15 and progress reports (2 pages) are due on March 20. Project reports are due at the end of the course final exam slot.
    • The students who choose the 12-unit section will work on a research project, with a research-style paper and an oral presentation delivered at the end of the semester. Ph.D. students may choose a paper format with the guidance of their advisors. For master students, a sample research-style paper can be found here: https://feifanginfo.files.wordpress.com/2016/11/2015_aaai_csworkshop_repeatedssg.pdf
    • The students who choose the 9-unit section will work on a systematic literature review or a project exploring the possibility of applying existing AI tools to a social good problem, with a survey paper or technical report and presentation delivered at the end of the semester.
      Sample survey paper: http://www.sigecom.org/exchanges/volume_15/1/FANG.pdf
      Sample technical report: https://www.aaai.org/ocs/index.php/SSS/SSS12/paper/view/4310/4644
    • Example project ideas: (1) Identifying Suicidal Thinking in Social Media; (2) Evaluating Predictive Models for Urban Crime; (3) Ensuring Fairness in Ridesharing Systems; (4) Deceptive planning.
    • Example literature review topics: (1) AI-based Prescriptive Analytics in Healthcare; (2) Understanding poverty through learning; (3) Principle-Agent Model and Its Applications

The grade will be calculated according to the following table (subject to change)

Course Component Weight Expected Workload Notes
Class participation 10% 3 hours/week 29 lectures in total. Get full score if attended at least 26 lectures.
Paper Summaries 20% 2 hours/week 14 paper reading assignments in total. The lowest score will be dropped.
Written Answers Assignment 20% 1 hour/week 8 written answers assignments in total. The lowest score will be dropped.
Final Project 50% 3 hours/week for 08-537, and 6 hours/week for 08-737 Report (30%) + Presentation (20%)