For those interested in pursuing careers in academia and research, graduate classes often serve as the first meaningful steps in learning the tricks of the trade, providing these students with a strong foundation that will help them identify efficient ways of critical thinking.
Unfortunately, going through such detailed classes for hours can be challenging, especially in the current day and age, when average attention spans are on a downward trend. In addition, relating the learned theory to real-life application is often difficult for theoretical classes, where the primary focus is on being precise in the endeavor. As a result, by the middle of the semester, students likely end up focusing on the grades rather than becoming familiar with the methodologies taught.
In my machine learning classes for graduate-level students, this can be amplified even more as the teaching begins on an advanced level, building on the assumed knowledge of mathematical foundations. Students seek a concrete list of steps, actions, and objectives that already exist in the literature rather than generating their own ideas. This can hinder investigative growth among students and make it challenging for them to develop out-of-the-box thinking.
I can incorporate research-oriented questions as a part of homework, but students tend to view them as a means to get points and often work on them on the eve of the deadline. In my opinion, this can disrupt the flow of holistic learning, and I wanted to explore how the class sessions themselves could do more to promote intuitive thinking beyond the materials being taught.
Based on some ideas I gathered from participating in the Kaneb Center Course Design Academy, I decided to incorporate a short brainstorming session into my spring 2025 class. After discussing with other faculty members and the Kaneb Center team, I realized it would be beneficial to add this right after I reach the 20-minute mark.
The outline of integrating the session into the regular classes is as follows:
- Start the class with a fresh topic or idea.
- Pose a related methodological question in the brainstorming session that I will cover in the second half of the class.
- Encourage participation and collect the answers. Connect the answers with materials to be presented in the second half and ensure the students understand them.
Because the question asked in the brainstorming session had not been previously discussed, the students had to rely on their imagination to propose a solution. There were no restrictions on the direction they wanted to take, but the focus was on using the ideas presented in the first half of the class to build their suggestions.
At the end of the semester, students mentioned that the brainstorming sessions helped them improve their confidence in taking such an advanced class. As most of the methodologies presented are intuitive, in almost every class, one of the students came up with a proposal that I had prepared to discuss in the second half of the class.
Award-winning works inspired most of the materials presented, and students were excited to see that ideas born out of their brainstorming sessions were remarkably similar to the state-of-the-art methods currently being used in the machine learning community.
Some students initially felt shy about proposing their ideas. However, they began to feel more confident as more of their classmates contributed suggestions. The brainstorming sessions also served as a great icebreaker, helping to foster interesting discussions among students from different departments.
Soham Jana is an assistant professor in the Department of Applied and Computational Mathematics and Statistics. He participated in the 2024–25 Kaneb Center Course Design Academy.