Spring 2011 PGM Class Evaluation Details

Post date: Feb 9, 2011 8:55:03 PM

We have decided upon the following evaluation scheme:

• 20% for class attendance and participation
• 40% for written reviews/critiques of two PGM papers (we will assign grades using your best 2 reviews out of 3)
  • 40% for presentation of a relevant research paper *OR* completion of a small PGM project

Review/Critiques: There will be three opportunities to write a review/critique of an assigned paper that uses PGMs for language and speech processing. Reviews will be due on the following Thursdays: March 10 (Approximate Inference paper), March 31 (CRF paper), and April 28 (Bayesian or Non-parametric Bayesian Models paper). The exact deadline is 24 hours before that week's lecture (to give us time to read the reviews and formulate feedback). You can write all three reviews if you like, we will take the best 2 reviews out of three for your evaluation.

You can choose to do either a presentation of a relevant research paper OR a small project using PGMs.

Presentation: The presentation will be a 10 minute class presentation on a PGM paper (not already reviewed/critiqued), followed by a class discussion period. Before you give the presentation, please schedule a meeting time with the instructors. At this meeting, you will give an advance version of your presentation and we will give you feedback. We will also ask questions to test your understanding of the material and help you resolve any issues. You can prepare notes for this meeting but no specific written summary will be required. Presentations must correspond to one of the three main course topics and will be presented in class on March 11, April 1, or April 29th, depending on whether you present a paper on approximate inference, CRFs, or Bayesian models, respectively. We will assign presentations based on your preferences but limit the number that can present on each topic.

Course project: For the course project, you can either do something from your own research/interests, or we can assign you a problem. The main goal is to simply get experience using existing PGM software to solve an NLP or Speech problem. The project will be formally assigned by March 1, and will be due on April 15th.

***Important***: By next Thursday, February 17th, please e-mail Shane at sbergsma@jhu.edu and indicate whether you'd:

A) Like to do a project or presentation

B) If a project, whether you have one in mind or whether you'd like us to assign you a problem

C) If a presentation, please rank the three topics from most to least interesting to you: Approximate Inference, CRFs, or Bayesian/Non-parametric Bayesian Models.

We may ask some presenters to switch topics if we have too many presenters on one topic.

Note that you are welcome to do both a presentation and a project, but please let us know in advance which one to include in your grade.

Our intention is to assign full grades for reasonable completion of each course component. That is, students that participate in class, complete two reviews, and make a reasonable effort at their presentation or project can expect to receive full grades (100%) in the course.