TL;DR: I tried blogging before and it didn’t work. I’m going to try blogging again and hopefully it’ll work.
I’ve always wanted to write a blog and indeed, at various points of time in my life, I did try to. Unfortunately, each and every one of those attempts at blogging fizzled out very quickly. But yet, here am I attempting to start a blog once again.
A reasonable person might ask what makes this attempt any different. While previously my motivation for starting a blog was one of “Everyone is doing it” or “I want to make a lot of money from serving ads”, that’s no longer the case now.
During my time at Berkeley, I was often very envious of older graduate students, who seemed to know a lot about a lot. As I prepare to enter graduate school in the Fall, it is becoming more and more evident to me just how little I know. A large part of this sensation is probably attributable to (1) the fact that I did my undergraduate degree in Mathematics but am entering graduate school in Computer Science (and therefore, am not as familiar with those concepts that my future peers may consider rather basic) and (2) impostor syndrome. Regardless, I still think the point broadly holds.
My hope is that this blog will be an effective excuse for me to learn about topics that I’ve been curious about but haven’t necessarily had the time or motivation to properly explore. There are two particular reasons why I think a blog is especially well suited for this task:
Learning through Teaching: During my undergraduate years at Berkeley, I found that the best way to learn was to teach. You simply could not teach, for example, probability or linear algebra to a group of students unless you had a truly foundational understanding of the material: they are often simply too bright and too inquisitive to let you get away with it otherwise. (This was especially true at Berkeley.)
Writing a (technical) blog is rather similar to teaching in that I believe the primary outcome of both is the dissemination of knowledge. I intend to write these blog posts in much the same way I would prepare to deliver a lecture at Berkeley: with a focus on clarity but without compromising rigor. And I hope that you, my (possibly imaginary) audience, will hold me to as high a standard as my wonderful former students did.
Open-Sourcing Knowledge: I am a big-believer in open-sourced knowledge. I cannot begin to count the number of times that I have relied on blog posts or YouTube videos to learn something after becoming frustrated by textbooks, which are far too often obfuscated by an abundance of symbols and jargon. I in fact fully expect that much of what I learn in the process of writing each blog post will itself have been learnt from a combination of other blog posts.
The open nature of a blog allows me to partake in this wonderful tradition of open-sourcing knowledge and to hopefully help share what little knowledge I have/ gain with those who seek it.
So, what can you as a reader expect from this blog?
As I previously mentioned, I hope to write about topics that I have been curious about, find interesting, and wish to share with others. This means that you can expect the vast majority of posts on this blog to be technical posts on topics that fall somewhere in the intersection of mathematics, statistics, and computer science; considering my background and research interests, you can especially expect a good number of posts on topics in AI/ML. To give you a better idea of what that means, I’m currently working on two posts: on (1) the multiple ways of looking at Poisson Processes and (2) autoencoders.
Additionally, I intend to write the occasional opinion piece, primarily to overcome my fear of taking a public stand on tough issues – a platform as open and permanent as a blog seems an appropriate place to confront that fear head on.
I’m aiming to update this blog somewhat frequently, with a new post every 2-4 weeks or so. I know this might seem rather vague but that’s intentional. I’m trying here to delicately balance (1) setting myself concrete targets to aim for and (2) giving myself enough room for error so I don’t beat myself up too much when I inevitably miss my target at some point :)