GPUs for Courses with Deep Learning

Project Title: GPUs for Courses with Deep Learning

Long Title (if desired): GPUs For Teaching Machine Learning, Computer Vision, Data Science

Project Lead's Name: John C. Femiani

Project Lead's Email:

Project Lead's Phone: 480-695-3199

Project Lead's Division: CEC

Primary Department: CSE 

List Departments Benefiting or Affected by this proposal:

  • CSE
  • ECE
  • Statistics / CADS
  • Potentially Biology / Bioinformatics students

Estimated Number of Under-Graduate students affected per year (should be number who will actually use solution, not just who is it available to): 70

Estimated Number of Graduate students affected per year (should be number who will actually use solution, not just who is it available to): 8

Describe the problem you are attempting to solve and your approach for solving that problem: This proposal will:

  1. Provide students (novice or upper-division) with immediate web-based access to Jupyterhub, and interactive online "executable paper" that is widely used in data science and has a demonstrated track record of classroom success.
  2. Allow instructors to give assignments and labs that require special computing hardware (GPUs), and it will enable students to share time on university or cloud-hosted GPU servers. Deep learning is an emerging and important topic that will remain important to a variety of data science and computer science problems. We have struggled to find cloud solutions and students have bought their own equipment in the past. Two courses would immediately be effected; others will likely also benefit.

How would you describe the innovation and/or the significance of your project: Students will be able to train larger models on realistic data sets, which is very important for learning. Faster training means students can experiments with different techniques to better understand how their decisions impact results.

Having local access to GPU servers means students do not need to compete with OSU or AWS bitcoin minors in order to have dedicated access to a GPU.

How will you assess the success of the project:

  1. Successful implementation and deployment to class
  2. Surveys of students -- which fraction if time is spend on dev-ops vs course outcomes. Expect to see a significant increase in time spend on course outcomes.
  3. Instructor assessment of the quality and complexity of labwork and projects. I expect to see an increase in the number of iterations; that is, in an iteration students will build a model, evaluate, form a hypothesis on how to improve it, and test the hypothesis.

Total Amount Requested: $27,310.00


Is this a multi-year request: No