Support of Data Science Course on Central HPC System

Dr Lim Chinghway, Department of Statistics & Data Science

Background

We conducted the course DSA4266 in AY2021/2022 Semester 1. This course requires the students to work on computer vision tasks, namely training and running state-of-the-art object detection models on both images and videos. We also expected the students to develop and host a Web UI for their solution. The computational resource requirements far exceeds the capability of most of the students’ personal devices and hence, we collaborated with the NUS HPC team and introduced the HPC as a platform for the students to develop, run and host their projects.

Solution

Throughout the whole duration of the course, we had dedicated HPC staff to respond promptly to the queries and requests of the module facilitators and students, ensuring that obstacles faced were resolved in less than 24 hours. For instance, the HPC staff were able to address our concerns regarding the contention of resources by the other HPC users  given the nature of the queue system; by setting up a specific queue that is to be used only by the members of DSA4266. In addition, they went beyond the expected help desk support as they took time off their busy schedule to attend the project presentation by the students, so as to help the students debug on the spot should any issues arise during their product demonstration.

Conclusion Remarks

This support model was perfect for our course and is likely to be so for the other modules alike. The support and flexibility provided by the HPC team ensures that the users are free to explore a variety of solutions without having to be concerned about the hardware and environmental constraints.

We thank the HPC team so much for all the support and help over these 13 weeks throughout the module period!