Learning Here, There, and Everywhere
Being in grad school, and in my case, a PhD program, entails an implicit requirement to maximize your learning. And I’d say that one of the most valuable skills are not necessarily the ones you possess currently, but the ones that you are able to possess. The best part of living in this day and age is that technology has substantially increased the knowledge that you are able to possess — no longer are you confined to learning from a single master as in medieval days — now you can learn from all sorts of “masters” and in a variety of different ways, from the physical to the virtual classroom, from the written blog to the podcast.
Learning is not just a hobby anymore, but has now become a career choice, and for better or worse, for those of us who want to continue advancing up the career ladder, it is a necessity for survival. As I progress further into the rabbit hole of grad school, it becomes more and more apparent that we simply can’t afford to stop learning when the work day is done. With that, comes the required break from having a “9 to 5” mentality, and in comes crazy attempts at starting days at 5 am with reading on immunology and genomics, listening to podcasts on machine learning during my commute, taking online courses at night on programming, and attending seminars and webinars when able I’m able to..
Of course, I’m not saying I do all of those in a single given day, or even week — unfortunately, I succumb to the limits of my mental bandwidth. And maybe the diversity is a symptom of trying to go too broad and not deep enough — the multi-taskers mentality. So, why (try to do) so many different things? Because I never know when or where that “spark” will hit. That “spark” could lead to the prototypical “aha!” moment, or could lead to beginning of a slow burn that culminates in clarity on my current work. Even when learning about things out of my depth, such as Latent Dirichlet allocation, its an interesting exercise to try to apply those lessons to areas of interest to you. For example, in the case of LDA, I could use it to assist me in generating models of cell-type specific gene expression, where the “words” may be specific genes, dynamic expression patterns, or modules.
In the spirit of sharing, I wanted to share some of the things I’m reading, listening to, or the combination thereof.
- Talking Machines : for anyone interested in learning about machine learning developments in the field (with some relevant background tossed in), listening to some great questions (how we define the “ultimate” goal for AI?), and interviews from some of the brightest minds in ML, this podcast makes machine learning digestible for a (slightly) broader audience. Basic knowledge of statistics helps get more out of it, but even without it, the explanations are intuitive enough that you’ll be learning without really trying.
- 99% Invisible : the first podcast that really got me hooked on podcasts, Roman Mars and crew are phenomenal story tellers that dig into the things that we ultimately take for granted, from the usage of revolving doors to the history of elevators or the story of the milk carton kids. Remember to always read the plaque!
- Freakonomics : the weird side of economics — having fallen in love with their book “Freakonomics” in high school, the podcast takes a similar tack to approach a variety of fun topics.
- Deep Learning (Udacity) : Google recently published a deep learning course on Udacity, and while I haven’t gotten to the meat of it (read: I’ve only looked into the intro), the study of how to use TensorFlow seems very promising with wide applicability to a variety of problems.
- Data Science Nanodegree (Udacity) : Stumbling out of Google’s Deep Learning Udacity course, I found that the relevant Data Science Nanodegree offers quite a lot, from databases with MongoDB to the fundamentals of data analysis in R.
- Coursera : my signature trick for Coursera: sign up for all the courses I can, and even if I don’t take any of them, I still have access to interesting courses down the line that I can peruse! Though finding the time for that is still a challenge in and of itself..
- Read by QxMD : for keeping up with the latest in literature, Read is second to none. I’ve tried to use RSS readers such as Feedly for this purpose, but Read does all the hard work for you, so you don’t have to fiddle with finding the right RSS feed for every single publisher you’re interested. Bonus, you can sign up for specific Keywords as well. This certainly beats Pubmeds email alerts/digest feature anyday! Plus, you can easily download the PDFs of papers in one-click (my favorite feature!), and then export the PDF to your favorite app (e.g., Papers, Mendeley). I am not overestimating how significantly this has changed my workflow for keeping up with the field — a must.
- Papers : for organizing all those papers that you’ll never read..extremely slick, though the lack of an Android app is a real sore point for me at the moment given that I bailed the iOS train (though it’ll probably be temporary at this rate, because not having Papers on my phone really, really sucks). Inserting references has never been easier. However, I will say that while Mendeley is slightly less slick, it has all the core features that make Papers great, so you really have no excuse to not be using one of the two.
- Feedly : for RSS, Feedly is second to none. Although I’ve essentially stopped using it because its annoying to configure your RSS..
- Medium : enter Medium — the “new” RSS. Medium is great for reading focused articles, rather than the endless sea of pointless status updates you might find yourself lost in otherwise on Facebook. Of course, it can’t compete with RSS in terms of sources, but if you’re looking for quick reads to fill in small gaps in the day, Medium can easily, and slickly, fit the bill.
- Twitter : follow all the scientists and developers, get timely updates on shiny new tools or technologies that are coming out. R bloggers is a great follow. Bonus: you can keep abreast of (scientific) celebrity feuds!
- Simplenote : I’ve tried Evernote in the past, but found that it was way too feature rich for my taste. Enter Simplenote: its the barest of bare minimums, with essentially a text only product, but its cross platform, has instant sharing via Tags (just tag someone using their email to share it with them, and they’ll get a link to the web app), and looks slick as well. [In case you haven’t noticed, I’m very superficial, and the UI/UX encompassing “slick” matters a lot to me!] It’s not without bugs, but overall I’ve found it to be much better than the bloated mess that is Evernote.
- Aeropress : smooth coffee, easy to make, and has a better tolerance for crappy grinds, unlike the French Press, which demands a coarse grind. Bonus: No bank busting espresso machine needed!
- Capresso Infinity Grinder : thank you mother-in-law-to-be!
- Tea : ever since getting loose leaf tea, I’ve loved tea so much more. Once you ditch the bags, you won’t be able to go back.