In the first quarter of 2023, I had the opportunity to delve into the world of Generative AI and work on S4/diffusion models for audio at Prof. Julian's lab, under the guidance of Hao-Wen Dong (https://hermandong.com/). This experience 1) allowed me to gain insights into ML for audio and learn about diffusion in genAI but more importantly 2) in general taught me how to setup and run ML experiments which has been a useful skill.

Here is a link to the project and samples; (link). Although I did not achieve the best results or publish a paper, the process was valuable in terms of understanding audio and S4, which I believe will be beneficial in the future.

Prof. Julian's lab was filled with brilliant individuals, and I decided to extend my stay for two more quarters, working on different projects and attending lab meetings and music reading sessions. This short stint opened my eyes to how difficult it is to conduct research and publish papers, and I gained a newfound respect for researchers who do so consistently.

Perhaps my tendency to constantly change my mind with new ideas and trying many things in the short time period of my MS program (5 quarters), which I prioritized over completion, resulted in many unfinished projects. However, this approach also allowed me to explore different areas of AI research, which is essential in the early stages of learning. This experience taught me about the importance of balancing exploration and exploitation in AI research. Exploration involves trying new approaches and ideas, while exploitation involves focusing on a specific approach to achieve optimal results. In my case, I learned I lean more towards exploration and I hope to continue doing so for many years to come.

During this period, I also took two courses: an Algorithms course with Russell Impagliazzo (https://cseweb.ucsd.edu/~russell/) and a Search and Optimization course with Sicun Gao (https://scungao.github.io/). In the Algorithms course, I collaborated with my teammates on a Wordle solving project (https://github.com/rohithaug/wordle-bot), which introduced me to the rigors of algorithms and the importance of definitions. In the Search and Optimization course, I had engaging conversations with Prof. Gao about AI and RL which led me to take a seminar with him in the following quarter and sit in for his lab meetings. (In this course I first learned about exploration vs exploitation!)