Abstract:

Objective:
This paper describes the experience and outcomes of delivering the machine learning course through project-based learning (PBL) pedagogy in a large class room of undergraduate engineering students in a private state university located in India. The findings of the paper provide a set of guidelines for engineering educators to implement PBL in the course, discusses the opportunities created and challenges faced by both faculty and students during the process.

Background:
Artificial Intelligence (AI) has become a one-stop solution to several problems. To equip students with the knowledge and skills required to provide AI-based solutions, a set of courses were designed that included 'Machine Learning', ' Deep learning', ' Computer Vision and Graphics', and projects including Institutional Research Projects, Capstone projects in the field. This paper's scope is limited to the experience in the course titled Machine Learning, a three-credit course that was offered to third-year electronics engineering students. The course content was delivered to 350 students by seven faculties, and the contents of the course focused on supervised, unsupervised, and self-supervised algorithms. The course used a bi-chronous mode of delivery and project-based learning pedagogy. Bi-chronous mode of delivery is when course content in pre-recorded video lectures is made available to students through moodle-based Learning Management System (LMS), followed by a synchronous class in the physical mode).

Research question:
The study qualitatively investigates the research question "What are the opportunities created and challenges faced by stakeholders when a machine learning course uses PBL pedagogy for the delivery?"

Methodology:
The paper describes the course execution pipeline that emphasized the phases including data generation, preprocessing of the data, and applying machine learning algorithms over the generated data to complete the given tasks, including regression and multi-class classification. The effectiveness of the pedagogy was evaluated qualitatively through the analysis of the interview data conducted for the faculty and purposively sampled students to note the opportunities, experiences, and challenges.

Results:
Data generation phase of the project resulted in several terabytes (TB) of data consisting of images of multiple classes and audio files which after processing, has a potential to emerge as a data set. The challenges faced by the students and faculty were related to operational issues such as time, increased workload, and limitations on the availability of resources like workstations. However, more themes have arrived through the thematic analysis of the interview data. Although the authors had the statistical data of the students' performance with and without the intervention of PBL, the comparison could not be made due to the confounding variable that the delivery mode of the course was different in the previous academic year.

Keywords:

Project Based Learning, Machine Learning, Artificial Intelligence.