Lately, I’ve been in awe of Scott H. Young, after reading his book, Ultralearning (which I highly recommend, by the way). He’s finished MIT OCW’s 4-year Computer Science curriculum in a year and has also traveled the world in different regions, reaching a near-fluency level in multiple languages. I get the idea that Scott is actually quite similar to me, except 1.) his determination and grit to succeed are impeccable, and 2.) he actually puts forth his plans. When he studied from MIT’s free resources (#respect, MIT. Really helps those tight on money and resources, here…), he studied straight from 6 am to 6 pm, with minimal breaks. When he pursued his language-learning ventures, he had promised and tried his best to only immerse himself in said language, and avoided English as much as possible to his ability.
The Problem of Instant Gratification
Frankly, I respect him a lot for these challenges. In a way, they’ve opened my eyes, as well. I’m glad to be born and living in the age of the Internet, where learning resources are vastly available everywhere. And let’s be real– a lot of us have explored sites like edX, Udacity, Coursera, Udemy, and other MOOCs, started our own course we were certain we’d finish with dedication, get to three videos, and then give up. Later, we tell ourselves. I think that’s the reality and the extent of most of our determinations, along with busywork from school, work, etc. distracting and taking away from our personal challenges. But what if… What if we minimized our ‘free’ time to be used on these skill-building courses? A little bit, every single day? In the long run, it’d make a massive difference, and I don’t doubt that. There’s a little bit of psychology, economics, and neuroscience involved in this entire fiasco as well– we as humans, we seek instant gratification, and truthfully, we actually do receive it in the real world.
From Amazon Prime to UberEats, or a couple Fortnite wins and the satisfaction from several likes on a photo on Facebook, they all are forms of gratification– to our brains, a rush of dopamine (the happy neurotransmitter). We are spoiled. Our patience has run out– any more than a week, and it’s become too long. We don’t see our marginal returns in the short run, although they’d come up in the long run. Some of us are motivated extrinsically when the real fuel lies within the intrinsic motivation to succeed.
An Honest Reflection
Honestly, I’m a victim of this attitude. A mix of it comes from over-involvement in school classes and extracurriculars, which really made my work-free-time balance topple over. The other half comes from overambition. Ambition isn’t necessarily a bad thing– it (for me, at least) serves as a model of intrinsic motivation, and it works. The issue arises when you set goals that seem interesting at first glance, then realizing it has prerequisites and then failing to address those very prereqs. It’s fun and great to look at your MOOC course list or library of books and think “wow, I’m such an awesome person– I have all these books and classes on chemistry, maths, and biology!” but it doesn’t necessarily mean that you’re great at all of these topics. Hell, it certainly holds true for me, somewhat. I have several books on these exact topics, and I enjoy reading them, but by far, I’m much stronger at humanities (unintentionally!) due to a casual habit of reading and writing I carried for years since I was a kid. I’d love to strengthen my base in all these STEM topics, and that’s why I’m about to pull a Scott and do this personal project.
I’m aware it’s going to take time, a lot of time. A lot of dedication, perseverance, and grit. A lot of intrinsic motivation, a lot of studying, and a lot of reading. That doesn’t mean I intend to turn my life into a prison of study– of course, I need PUBG, Reddit, Twitter and YouTube in my life– but I think setting a time-wise goal will help me. I’ve spent the past few days compiling resources (similar to packing my clothes and essentials for a backpacking trip abroad) in a framework schedule. I think I’m ready to embark on my journey. I’ll also write updates accordingly. Now, let me get to the main part– what exactly am I challenging myself to do?
The Challenge
Technically, it’s two challenges, but for stylistic purposes, ‘The Challenge’ just sounds a lot cooler than in plural form. I’m overthinking it again. Moving on.
The first is to get a strong base in Artificial Intelligence and any subsets I find interesting along my journey, whether it be NLP, ML, or deep learning. I’ve worked on surface-level ‘research’ that involved VR or AI, but I’ve never gotten the chance to really dig my fingers into it. In my opinion, it wasn’t really research. Thus, this challenge. If I couldn’t achieve what I wanted from my past experiences, I will get it myself. This is a year-long challenge, and I aim to get graduate-level knowledge on these topics by the end of it. That’s right. Graduate-level. Meaning, on September 3, 2020, I should be there– or at the very least, close to it.
Phase 1 — I intend to ACTUALLY work on prerequisites. In the past, because I sought instant gratification, I jumped straight into ML and AI, and quit pretty fast as you know, possibly also because I had a weak conceptual and mathematical understanding. Fortunately, I’m taking multivariable calculus at college, so I’ll have that plan of study to speak for itself, and I have a basic statistics background, but I lack the background in Matrix and Linear Algebra, which is the backbone of neural networks and ML. Because AI and its subsets also require advanced statistics, I will also be studying that. I expect a good mastery to take anywhere from 2-3 months, but I could try to quicken the pace. I may also take the corresponding courses in the Spring semester to solidify and cement the knowledge, especially if I go through with adding a CS minor/major to my neuroscience major. I do have an alright base in programming, but I think I could really strengthen Python and TensorFlow, Keras, etc. I might take ‘Intro to C’ in Spring sem, so once again, the second semester might shape itself.
Phase 2 — Jumping into the MOOCs, MIT OCW courses on AI, ML, and advanced courses. Now that I have the prerequisites, I can go into the basics on the topics themselves, reading relevant books/textbooks, and also improving and reinforcing my programming skills. Hopefully, if I’m involved in research by this time, it’ll allow me to apply my knowledge and also learn practical usages in the lab. I’d also like to attend Hackathons. I’m pretty pumped for this phase, to be truthful.
Phase 3 — Now, into more advanced, graduate-level coursework on AI, ML, and the specialty I’ve found fascinating. Hopefully, more involved hands-on in research, Hackathons, and reading and interpreting a lot of research papers in the field. Ideally, I’d also like to start testing solo projects for a better, stronger feel. My framework for this phase isn’t as strong as the others, but I think it’ll reinforce itself over time. And alas, that is it.
The MCAT Challenge
The other ‘challenge’ project isn’t as elaborate as the AI one, because it involves what I already study– however, I would also like to somewhat repeat this same project, except in the fields of Bio, Physics, Chem, and Biochem ideally. Although, I’m not too worried about this– I will be having the courses needed for mastery throughout college, regardless. However, I’d like to get a 520+ on the MCAT, and that’s going to be tough without a strong background. I intend to use MIT OCW to supplement my regular college classes, and I’m developing a framework for deadlines to finish certain MCAT books (for eg., I’m working on the Kaplan Gen Chem one at the moment, and I’d ideally like to finish by the end of this week, etc.) This project is mainly for reinforcement.
That being said, these are the only challenges for now. Although I’d love to learn Mandarin Chinese or Korean in depth by traveling, that venture can wait a bit.
Determining Success or Failure
First of all, I want to make it clear that there isn’t actually any real failure in the end result. Regardless of the level I attain, I have gained knowledge in one way or another, and that is a success to me. For both the AI and project– there isn’t really much room to determine the level of understanding or the level achieved besides perhaps taking timed final exams from OCW or other resources. That being said, I think these are minor details I can figure out with time, as mentioned before.
Final Thoughts
I’m a current college student. I think I’m a little fortunate to be starting this journey early, as it’ll help me in the future in terms of careers and help steer the direction I want to engineer my life, however, I also hold the tenet that age simply doesn’t matter when it comes to our goals, in the end. I believe that anybody is capable of doing the challenges Scott did with proper discipline and intrinsic motivation. These challenges don’t have to be limited to academia, nor do they have to be incredibly complicated or difficult– it could be fitness, languages, small chores, and tasks, and even reading! The world is waiting for all of you to do something great with your life– do it!
My Sources and Materials
That being said, I intend to build up a list of the resources that I use for my projects. I’ll develop one for the MCAT in a separate post (as most resources are TPR, TBR, Kaplan, EK, etc. books), but for online AI resources, here’s what I have so far– more to be added with time. (NOTE: Reddit links tend to have a plethora of awesome information on each post)
- http://neuralnetworksanddeeplearning.com
- https://www.reddit.com/r/artificial/comments/59iff5/how_best_to_use_1_intense_year_to_learn_ai/
- https://www.reddit.com/r/artificial/comments/5kf6l7/how_can_i_begin_to_learn_the_basics_of_ai/
- https://www.reddit.com/r/learnmachinelearning/comments/adwft2/all_the_math_you_might_need_for_machine_learning/
- https://www.reddit.com/r/learnmachinelearning/comments/9uxq13/best_statistics_books_for_machine_learning/
- https://howicodestuff.github.io/machine_learning/2018/01/12/a-roadmap-to-machine-learning.html
- https://www.kadenze.com/courses/creative-applications-of-deep-learning-with-tensorflow-iv
- https://towardsdatascience.com/linear-algebra-for-deep-learning-f21d7e7d7f23
- https://youtu.be/5v1JnYv_yWs
- https://www.youtube.com/playlist?list=PLtmWHNX-gukIc92m1K0P6bIOnZb-mg0hY
- Udacity courses on Machine Learning, Tensorflow, Intro to Artificial Intelligence
- https://mithi.github.io/deep-blueberry/
- https://www.reddit.com/r/artificial/comments/89zau1/how_do_i_learn_artificial_intelligence/
- https://github.com/owainlewis/awesome-artificial-intelligence
- Andrew Ng – Machine Learning (Coursera) and other relevant courses on MOOCs
- MIT OCW curriculum for CS
Challenge officially on!
Neurologic 