Essentials of getting into Machine Learning
You may have probably seen Robert Downey Jr.’s YouTube Original docu-series The Age of A.I (If not, it’s a must-have on your watch list) and just thought how bonkers the applications of A.I could be. For me, it served as a push off the edge into Machine Learning after a year and a half in product management in a software development start-up.
Frankly, I’m embarrassed to admit that I didn’t possess the state of mind to think I can actually drive change or play a part in the ML space and being content as an enthusiast and really a spectator of the space watching the amazing feats being realized(You’ve probably read one of those AI-buzzword filled articles on Techcrunch) and like a kid wanting to throw hoops on the floor, just yearning to be the guy in the middle of the action. I can probably blame so many things here including my circumstances but the truth is I have everything I need to get into it(a good laptop, internet, a decent math background and a healthy dose of commitment and discipline).
Why the post though?
“People don’t buy what you do; they buy why you do it. And what you do simply proves what you believe”-Simon Sinek, Start With Why.
This post is in many ways a culmination of a year-long passion as a machine learning enthusiast and observing how data can be optimized to have the most impact in businesses. I hope to lay out a plan that I will commit to for the next year putting in around three hours every week to grow to be a machine learning engineer. I am really convinced that Machine Learning(AI) will be in the crux of the 4th Industrial Revolution.
My reasons for doing this guide:
- I want to stay accountable and post monthly updates(or every other month depending on my schedule) on what I have been learning. This would also be a good point of reference to track my growth.
- I would love to coalesce with a bunch of crazy folks who – your mission should you choose to accept- are interested in Machine Learning and would be willing to step into the learning curve alongside me(We will have a virtual classroom bouncing ideas off of each other)
- I am interested in working on a couple of ML algorithms that I would love to employ in a year’s time: some recommender and supply chain optimization algorithms.
What’s AI/ML(A simplified explanation)?
The goal of Artificial Intelligence is simulating any mental task. It’s a really simple statement but extremely loaded. Its subfields include search and planning(e.g AI playing Chess), reasoning and knowledge representation, perception(computer vision), the ability to move and manipulate objects(robotics), natural language processing and machine learning. This guide focuses on the subfield of Machine Learning which really has its applications in every other field of AI. Machine Learning’s goal is basically pattern recognition: It observes a pattern, makes a prediction and updates the learning.
Without getting bogged down with the details(There’s a slew of articles addressing this topic), the podcast Machine Learning Guide by Tyler Renelle was the best lay-of-the-land tutorial and primer through the details of Machine Learning(a must-listen for anyone really interested in this space and would like to follow through with this guide). If you come from any SMET background, a lot of the math you learnt in school may have felt overwhelming or really non-contextual. Almost like learning about trees in a forest and never knowing how the forest looks like. This podcast (especially Episodes 1–10) helped contextualize an elevated perspective of Machine Learning and even as we get into the Mathematics, there’s a greater appreciation for the algorithms studied. It also helped me curate the essentials of what to study.
The Essentials Guide
Getting into it; I have boiled down the essentials to four parts that are integral to see one as an ML practitioner. I have planned it as a year-long guide (3 months on each part with roughly 4 hours/week study time). It can probably be done in a much shorter period, these are my reasonable estimates for a busy/working individual.
The learning schedule has an 80–20 % split between the Machine Learning aspects and Mathematics aspects involved.
For the Machine Learning Resources, We will chronologically do:
- Andrew Ng’s Coursera Course: Machine Learning by Stanford University. The most fundamental course to begin from. Estimated time of completion(3 months)-Free
- Deep Learning Book by deeplearningbook.org with associated videos(Estimated time-3 months)-Free
- Python(Est. time-2/3 months)-Free
- Tensorflow from the documentation from their website and some relevant video tutorials(Estimated time 2/3 months)-Free
For the Mathematics side, we shall primarily use Khan Academy(Free)(This is optional and you can pretty much come up with your own schedule here depending on how confident you are in your Math skills)
- Linear Algebra
- Statistics(Probability and Inference)
- Calculus 1,2 & 3
There you go folks. Excited to jump into this ride, welcome aboard.
If you have some audio time (during commutes or chores), there are some awesome resources that you can supplement your learning. These resources are optional but extremely cool 🙂 – This could be an awesome segway into trying audible or getting into podcasts:
- The Singularity is Near by Ray Kurzweil
- The Master Algorithm by Pedro Domingos
- The Machine Learning Guide Podcast by Tyler Renelle
- O’Reilly Data Show Podcast
If you’d want to jump into this bootstrapped course with me, feel free to ping me on Twitter!