Information About 16.420 / The Graduate Version of this Subject
Please Log In for full access to the web site.
Note that this link will take you to an external site (https://shimmer.mit.edu) to authenticate, and then you will be redirected back to this page.
Course Description
An introduction to representations and algorithms for artificial intelligence. Topics covered include: constraint satisfaction in discrete and continuous problems, logical representation and inference, Monte Carlo tree search, probabilistic graphical models and inference, planning in discrete and continuous deterministic and probabilistic models including MDPs and POMDPs.
Graduate student section
Students enrolled in the graduate subject, 16.420, will be required to complete additional assignments to differentiate levels. These additional assignments will consist of the following:
- Three mini-projects
- One research paper summary presentation
Paper presentations
Each student in 16.420 will be required to give a 30 minute presentation on a research paper related to the material in this subject. The paper presentation will be 10% of the final grade.
The research paper summary presentations will be held on Fridays from 10-11am in 33-418 starting on Friday, October 6th with an information session, and the presentations to start on October 13th. Attendance at the presentations is not mandatory but we will assign two additional discussants to each paper and ask the discussants to attend their assigned presentation.
The paper presentation schedule is available here.
Note that one presentation will occur on a Thursday, November 9 to accommodate student schedules and the institute holiday on November 10.
Additional guidance is contained in these slides that were presented in the information session on October 6.
Grading for the paper presentations
The presentations will be graded according to the following rubric:
- Introduction and problem statement (10%)
- Related work / placing the paper in context (20%)
- Summary of approach (10%)
- Experimental results (10%)
- Analysis of results and limitations (10%)
- Results that followed (or could follow) after the paper was published (20%).
The structure of this grading scheme can be considered as a template for the presentation. Notice that 40% of the grade is for content not contained directly in the paper itself, but contained in related work and papers that may have come after the paper itself that explain the impact of the paper. If you have difficult finding follow-on results due to the recency of the paper, you can also discuss your assessment of what the follow-on results and the impact are likely to be.
Each student will also be assigned a role as a discussant on other papers, and will receive 20% of their presentation grade for their engagement on those other papers as well.