By Dr. Sourya Dey
I’m here to talk about the thing known as a PhD. What can you expect once you enroll in one? Is it worth spending 5-6 years of your life researching on new stuff? Well, let me try to answer these questions on the basis of my own PhD journey, which I recently completed.
A PhD is not a box to be ticked off, like something you have to get off your to-do list. Neither is it simply a prelude to landing a six-figure job. Instead, think of a PhD as a way of life. Just like life itself, a PhD does not turn out in the way you expect it to. You might have seen those memes depicting the contrast between ‘Expectation’ and ‘Reality’. That applies to your PhD years as well. Here’s a little map to show what I mean (or, as a PhD would say, refer to Figure 1):
Why am I telling you this? Haven’t you gone to all those academic seminars which chart out the perfect PhD course where you are guaranteed to succeed after 5 years? Well, the reality is, things don’t usually turn out to be so straightforward. If you are about to end your PhD and you feel everything has been smooth, I would seriously doubt whether you actually did a meaningful PhD. You see, a PhD is a journey into uncharted territory. If you set out to explore undiscovered land, you would not expect that place to already have a hotel and a pizza place. Similarly, don’t expect your PhD to proceed in the ideal fashion. Instead, ride the waves. That’s what makes it so wonderful!
When I started my PhD in 2014, I knew that I would become a circuit designer. I was all excited about fabricating chips and took the relevant courses in my first year. It was only then that I realized I didn’t like the field of circuits. Now don’t get me wrong, circuits is an excellent field with great research potential. But it was not for me. Then what was in store for me? At the time, I didn't exactly know anymore. You can imagine the kind of stress that might put on a young researcher. What I did know was that I had come to USC to explore stuff, not to graduate as fast as possible. And so, after one and a half years and writing a journal paper, I quit from circuits.
I joined Prof. Peter Beerel’s group in January 2016 and he offered me a project on superconductivity. Sounds pretty cool, right? Unfortunately, I had absolutely no idea how anything worked. Why did I take it up? Well, I didn’t know what else to do. Two months later, during spring break 2016, I received a call from Peter while on holiday in DC. He told me about this new project on machine learning where he was planning to collaborate with Prof. Keith Chugg, and asked if I wanted to work on it. If I agreed, I would become the student to kickstart this effort. Yet again I was in a situation where I had no idea what my professor was talking about. This might sound ludicrous now, but this was in 2016 and machine learning hadn’t yet exploded the way it has now. In fact, Peter and Keith were also in the same boat as me. Machine learning was something new for all of us. And so a new research group was born, which later came to be known as Hardware Accelerated Learning - or HAL.
This is a point I would like to highlight about USC Viterbi engineering in general – there is a heavy focus on evolving with the times and staying well-connected with the most exciting problems in the field.
The early days were challenging. We did not have the funding, resources or knowhow to compete with the big players in machine learning. However, it was a lot of fun. Being a nascent group meant that a lot of our focus was on experimentation and exploration. This suited me beautifully since I could work according to my own style and flair instead of following a heavily trodden path. The lack of funding meant that I worked as a teaching assistant multiple times, which has instilled in me such a love for teaching that my ultimate goal is to become a professor. I am also proud of the impact my work made, not necessarily in terms of number of citations and other bean counters, but more on the lines of building a group up, guiding a new venture, and eventually securing funding in 2018.
Peter and Keith were terrific advisors to me. I am quite amazed at the way they took up a totally new field at a mature point in their careers and excelled in it. In fact, Keith has gone on to teach the first deep learning course in the ECE department. This is a point I would like to highlight about USC engineering in general – there is a heavy focus on evolving with the times and staying well-connected with the most exciting problems in the field. What I especially like is the inter-disciplinary nature of research, for example, Prof. Leana Golubchik and her students from computer science are an integral part of HAL. It is with fondness that I look at HAL’s journey from 2016 to 2020; the group currently engages three professors, six PhD students, a post-doctoral researcher, several B.S. and M.S. students on a rolling basis, and collaborates with other professors at USC and ISI. The group receives funding from multiple agencies, has published 15 papers and several posters at various conferences and journals, and won awards for Best Paper and Best Poster. The future looks bright, but if there’s one thing my PhD journey has taught me, things are never straightforward!
All of which brings me back to the initial cartoon on PhD ‘expectation’ vs ‘reality’. The journey twists and turns, peaks and falls. There are times when you don’t want to work on something and times when you have no idea what’s going on. There comes a period (or periods) when everything looks dark – aptly described as the ‘Valley of S**t’ in this article. But there are also times when you feel you have done something truly worthwhile. Maybe it is the way you approached a problem. Or the way you solved an issue, even if it was something small. Maybe it was a smart piece of code that you wrote. It all helps you to grow as a person. At the end of the day, whether it takes you five, six, or however many years, you will be glad you undertook the PhD journey. Just as you are glad that you are alive, even during these times of Covid-19. To put it simply, PhDs rock!
Dr. Sourya Dey Graduated from the Ming Hsieh Department of Electrical and Computer Engineering in May, 2020 with a PhD in machine learning. To see more about him and his research, visit his website.