In the upcoming semester, I will be continuing three different research projects and will have openings for a few students who wish to pursue them for ISE-590 Directed Research credit for each project. The expectation for this credit is that the student averages 3 hours per week per credit (thus, 6 hours per week if 2 credits are requested). Each student will be given specific objectives at the beginning of the semester and will be expected to produce a report at the end of the semester documenting their results. Note that ISE-590 is only given on a pass/fail basis.
As I have a full teaching load this coming semester, the teams will need to be able to work with minimal supervision. My plan is to meet with each team in person for one hour per week on Wednesdays on campus and to name one or more “research team leads” to help me coordinate the activities of the various team members.
Please note that these projects are not pure research in that they are not attempting to develop new methodologies or theories but instead are very applied and are focused on learning about existing analytics techniques and analyzing and documenting their effectiveness and utility.
Automated Dataset Generation and Grading
Continue this ongoing project with the following goals for the fall semester:
- Complete initial functionality, test its use with a “live” class, and support the generation of a paper for publication at the ASEE 2024 national conference
- Perform tasks related to preparing the new tools for publishing and distribution to other faculty members:
- Create document for future TAs on how to set up an autograded assignment with randomized individual datasets
- Complete and test new front-end user interface for autograding
- Package functionality in Github
- Perform research to identify additional functionality for future releases.
- More flexible auto-grading capabilities
- Enhanced capabilities for specifying and generating different types of datasets.
Cloud-Based Analytics Platforms
For the spring semester, I am combining the two projects formerly referred to as Automated Feature Engineering and Model Ops into a single project that more generally researches and documents cloud-based analytics platforms.
The overall objectives of this project are to research and document the various cloud analytics platforms currently on the market with the objective of creating a presentation to be used as part of my ISE-543 class in the spring semester and, ultimately, to serve as an overview/survey paper on the topic including a reference architecture that generalizes the capabilities of cloud-based analytics platforms.
The team will break into small sub-groups to research several of the current products and, to the extent practical, to develop prototype model pipelines that include demonstration and evaluation of the following capabilities:
- Cloud-based data stores
- Feature engineering and data pipelines
- Model development and training capabilities
- Model registries
- Model deployment
- Model monitoring
- Model updates and support in production