Toward Scalable Laboratories in Signals and Systems: Content, Deployment, and Grading

Three years ago, we introduced Jupyter Notebook labs in our upper-division course in Signals and Systems at UC Berkeley. In this paper, we highlight three of our recent innovations that significantly improved both student and instructor experience. We’ve designed five industry-inspired labs that appeal to our students’ interests across a broad spectrum of application areas. We showcase one of them here. We’ve adopted DataHub, a cloud-computing platform to deploy a common portfolio of scalable resources that enable students to work on their labs from a browser-based environment remotely. This eliminates their erstwhile worry about messy Python package management or the limits of their own computational resources. And we’ve devised a method to autograde lab assignments, which emancipates instructors from the constraints of manual grading. In an era of limited staff bandwidth, a salient benefit of these innovations is substantial scalability in class-enrollment size.