Wednesday, 18 May 2016

But really, why San Diego?

Choosing a grad university and advisor should be given about one-fourths* as much importance as choosing one's life partner. It's a significant decision since it involves investments of thousands of hours in time, the study happens in one of the most exciting periods of one's life, it's also the last and arguably the most significant period of focused study.

With this in mind, perhaps the last post ended too quickly. Now that enough time has passed, I can, from my current vantage point, present the big picture of my decision making. Hopefully, a picture that's big (or small?) enough to be interesting and yet covering all the essential details. Very often, coming up with this 'right' level of abstraction is quite the challenge.

This is a goal that's sought after in all scientific disciplines, quoting Einstien "Everything must be made as simple as possible, but not one bit simpler". I've always had an interest in such simple, elegant and fundamental ideas, with repercussions that are wide and closely coupled to our everyday world. While in high school, this made me develop a special interest in physics. Concurrently, I also developed a certain disinterest in mathematics since all it seemed to bring to the table was a set of tools, but with no direct repercussions.

However, these impressions were reduced to 'bits' as I came across Shannon's information theory in college. The theory, which quantified the all-pervasive notion of information into bits (0s and 1s) and which helps understand the fundamental limits of compression and reliable transmission of information, was simply marvelous. As I explored further, in certain problems, I also came up with strategies and algorithms which attain the holy grail - the information theoretic limits. I found this to be an exercise like no other and thus, through information theory, I discovered just how closely coupled to everyday life fundamental notions derived purely from mathematics can be! I also realized that my real interest is in cracking questions which are more fundamental than applied.

Now, over the course of my studies, it's the boundary between CS and EE that I've found to be the most interesting. Algorithms, learning theory, compression, coding theory and estimation (aside from information theory) are topics that I liked the most. Of these, learning theory is of special interest to many lately. With the large quantity of data and computational power at disposal, it's of interest to a large number of people about what can be inferred from this data. Many researchers from traditional disciplines have been shifting to this area, carried away by this wave of funding interest.

And since it's always a good idea to follow the trending areas of funding interest, when I identified a certain group which works on fundamental problems in statistics and learning theory and who have been very successful in their research, with the advisor having special interest in my most favorite part of information theory - compression, and with an excellent track record of graduating students in the group, no further questions remained! Needless to say, I'm very excited about what lies ahead!

*Might correct this a decade later; will be interesting to know if I was right about this one.