CSE 481N: Natural Language Processing Capstone

University of Washington

Spring 2019

The syllabus is subject to change; always get the latest version from the class website.
Meetings:MGH 058, Tuesdays and Thursdays 10:00–11:20 am
Poster session:Tuesday, June 11, 12:30–2:30 pm, Allen Center atrium
Instructor:Noah A. Smith (nasmith@cs.washington.edu)
Instructor office hours:by appointment, CSE 532
Teaching assistants:Elizabeth Clark
Lucy Lin

1 Course Objectives

This class will provide students with an intensive 10-week experience in successfully completing a challenging, well-scoped research project.

Participants will work in small groups (approximately 3 people in each group) to hone their technical skills to quickly absorb and adapt new technical knowledge, gain experience in complex programming, perform thorough experiments and analysis, and learn how to find a path when faced with negative results.

Additional objectives of this class include:

technical communication skills to produce high quality interim technical reports that inspire insightful discussion across project groups,
advisory project experience to provide technical advice and constructive feedback to others, and
project management skills to prioritize work items to maximize the chance for successful outcome.

2 Milestones and Artifacts

By each of the following dates, you will make a blog post that describes exciting progress (new trials, results, failure, insights) your team has made. You will submit to Canvas the URL to your new blog post each week. Keep in mind the overall milestones when planning out your project activities.

List of blog posts (those in boldface carry extra weight in the final grade):

Due April 9: team name, list of members, top three project ideas you’re excited about, minimal viable action plan for each with stretch goals, start a github project (share URL)
Due April 11: establish pros and cons of each project idea (consider especially whether you have the resources to succeed, how excited you are about the project, and whether a successful result would be useful to the world), optionally narrow to one or two; establish likely codebases/platforms, get them installed and become familiar with them; identify topic on which you’d like a lecture or class discussion
Due April 16: project proposal. Include both the minimal viable action plan as well as stretch goals. Clearly state your motivation, related work (literature survey), project objectives, proposed methodologies, available resources, and the evaluation plan.
Due April 18: complete at least one strawman/baseline approach, run experiments, and set up the evaluation framework.
Due April 25: complete multiple strawman/baseline approaches, record their performance, plus perform error analysis.
Due May 2: first advanced solution attempt. What did you try? Are there any exciting results? Any confusing results? What are the failure modes? What will you try next?
Due May 9: continue advanced solution attempt #1, run more experiments, do more error analysis, and sketch out the next action plan. Your blog post should include a summary of the group feedback discussion that you had.
Due May 16: second advanced solution attempt. Similar to blog post #6.
Due May 23: continue (similar to blog post #7).
Due June 6: final report. Details on the format and structure of the final report can be found in this latex source file, which has this accompanying bibtex file.

Your team is also expected to give five five-minute updates over the course of the quarter (between April 11 and May 30). You should sign up for these in advance (contact the TAs); they should be spaced out over the quarter and we will only have four presentations per class meeting. You are required to use no more than three slides (including the title slide) for each update.

The course culminates in a poster session, held on Tuesday, June 11, at 12:30, in the Allen Center atrium.

3 Evaluation

Students will be evaluated as follows:

Your team may use up to three late days for blog posts (but not for the project proposal or final report). Feedback to other teams will only count if it is provided within twenty-four hours of their blog post or in-class presentation.

4 Computing Resources

CSE has reserved GPUs nlpg00 through nlpg03 for you to use for this course. See here for information about other resources.