A self-learning guide for anyone to get into Artificial Intelligence

Original article was published on Deep Learning on Medium

No, that is not a click-bait title. The beauty behind the field of Artificial Intelligence is that it is not limited to mathematicians and computer scientists anymore. Of course, it should have been clear to us in 1958 when Rosenblatt first gave us the Perceptron model; we should have known that we shouldn’t limit research in Artificial Intelligence to mathematicians and computer scientists. We’ve had a 50-year delay, but that is fine, let’s move forward.

Modern Artificial Intelligence is not limited to mathematicians and programmers; it shouldn’t be. Me and my peers at PhospheneAI have always believed and pushed people from other fields to get into this field called Artificial Intelligence; we’ve recently started pushing and investing our efforts in pulling economists and psychologists into this brilliant field of Artificial Intelligence. Also, our entire base in PhospheneAI and our work is largely inspired by Computational Cognitive Neuroscience, which in itself a conglomerate of fields.

We are just scratching the surface. Neuroscience, Psychology and Economics are fields where we at PhospheneAI found potential applications and we’ve found new potential in Human Resource Management as well. And as I said, we are scratching the surface. We haven’t explored too many fields like pharmaceuticals, tele-communications, marketing, fashion, film and advertising, civil engineering, material science, etc. There isn’t a field that I can think of that Artificial Intelligence cannot intrude and produce groundbreaking innovations.

This is the reason why we believe that Artificial Intelligence cannot be and should not be dominated by computer scientists and mathematicians. We do have a role in advancing the field, but that alone isn’t enough. Plus, we computer scientists have a history with messing things up; place a bunch of us in a room and we kind of become morons (LOL).

So irrespective of what field you’ve graduated in; irrespective of what your age is or even if you are retired; irrespective of whether you’ve done computer programming or not; irrespective of whether you like mathematics or not, if you feel excited by the word Artificial Intelligence, you must get into this field. We need people from other fields to contribute to this massive field and help build new products together.

This article is a guide for anyone, with little or zero knowledge in computer programming, statistics, linear algebra and calculus to start getting their feet wet in Artificial Intelligence. Now, remember my passion for Artificial Intelligence didn’t come from my passion for programming and mathematics. In fact, I was never a huge fan of mathematics. My accidentally discovered passion for Computational Cognitive Neuroscience is what drove me into falling in love with Calculus, Linear Algebra and Statistics. So, don’t worry if math scares you now, once you get into a field where you can apply your existing passion in mathematics, you’ll fall in love with it too.

So, let’s begin with the journey. Every time, I mention a course, I’ll place a link following it. So, don’t get confused with that. The links redirect you to online learning platforms where you can access the courses that I am talking about.

Step 1: Understanding the big picture

I believe in understanding the big picture before diving deep into something. So, the first course that you should probably taking is a non-technical course that explains the current state of AI and it future. And for this, there is no better course than Andrew Ng’s ‘AI for Everyone’ course on Coursera.


Step 2: Getting your feet wet in computer science

Once, you are done with the above course, you’ll need to get started with programming. Now trust me, computer programming isn’t as scary as people claim it is. It is just another language, in fact, it is much easier than learning a natural language like Latin or French or Sanskrit. Programming languages have a strict syntax and there are very generic rules, which isn’t present in natural languages. So, programming languages are extremely easy to learn. Now, there are plenty of introductory programming courses, but there is this one course from Harvard University which as happened to be one of the first MOOCs that I took. The course is named CS50 by Harvard University.


CS50 is taught with non-computer scientists in mind, meaning, the course is intended for people from other branches and fields. In fact, the instructor David J Malan, himself wasn’t a computer science student, but rather took Latin history and Dramatic Arts in his undergraduate degree. So, there is no better course to begin your journey in computer science.

Step 3: Mathematics — The Language of Nature

Now, parts of this step are optional. Mathematics is extremely important in Artificial Intelligence. But, to get started with Artificial Intelligence, you don’t have to be a math expert, however, you definitely should be comfortable with a small portion of linear algebra and calculus. Now, remember, the way we were taught Calculus and Linear Algebra were completely different. Remember how you were asked to memorize formulas in Differentiation and Integration and crunch a number of problems in your textbook?

That isn’t the case here. In real life, it is not important to be able to work out complex problems, but rather you only need to understand what differentiation actually means in real life. You need to understand why calculus was invented and how it easily explains how you can What does differentiating a function with respect to a variable actually means is important than having to solve complex problems. I’ll list the resources below and take them in chronological order and if you want, you can ignore the optional ones, but I do recommend that you take those too.

3.1 Essence of Linear Algebra — https://www.youtube.com/watch?v=fNk_zzaMoSs&list=PLZHQObOWTQDPD3MizzM2xVFitgF8hE_ab

3.2 Essence of Calculus — https://www.youtube.com/watch?v=WUvTyaaNkzM&list=PLZHQObOWTQDMsr9K-rj53DwVRMYO3t5Yr

3.3 Statistics with R — https://www.coursera.org/specializations/statistics

3.4 (Optional) Mathematics for Machine Learning — https://www.coursera.org/specializations/mathematics-machine-learning

3.5 (Optional) The Matrix Calculus you need for Deep Learning — https://arxiv.org/pdf/1802.01528

Step 4: Machine Learning

Alright, lets dive into the real stuff now. Now, when anyone talks about using AI in their products, they are talking about using a specific branch of AI called Pattern Recognition. I cannot think of a product that claims to use AI but doesn’t involve Pattern Recognition. Another fancy term for Pattern Recognition is called Machine Learning. Machine Learning is the process allowing a computer to understand the real-world using data. Let’s say we want a computer to perform facial recognition of all the people in a company. We show pictures of various people in the company to a computer program, which automatically learns how to identify each person. It sounds like magic, but trust me, it is just some Linear Algebra and Calculus. This process is known as Machine Learning.

There are too many good sources to learn machine learning, but here are my favourites. Take them in the same order.

4.1 The open Machine Learning Course: https://mlcourse.ai/

4.2 Machine Learning from Stanford University: https://www.coursera.org/learn/machine-learning

4.3 (Optional) Stanford CS229: https://www.youtube.com/playlist?list=PLoROMvodv4rMiGQp3WXShtMGgzqpfVfbU

Step 5: Deep Learning

Machine Learning is a broad field and there is a Machine Learning algorithm known as Neural Networks; Neural Networks are partly inspired by the working of Neurons and the brain. Since 2012, the innovations made in Neural Networks has been tremendous that almost every image processing and speech recognition system used today is powered by Neural Networks. If you use large scale Neural Networks to understand data, then it is known as Deep Learning. Machine Learning is the backbone of Deep Learning, so if you understand Machine Learning, you shouldn’t have trouble getting into Deep Learning. I’ll list the resources below, take them in the same order.

5.1 Deep Learning Specialization: https://www.coursera.org/specializations/deep-learning?

5.2 CS231n from Stanford University: https://www.youtube.com/playlist?list=PLkt2uSq6rBVctENoVBg1TpCC7OQi31AlC

5.3 Courses from fastai: https://www.course.fast.ai

5.4 Deep Learning from New York University: https://www.youtube.com/playlist?list=PLLHTzKZzVU9eaEyErdV26ikyolxOsz6mq

That’s it! I know that it is quite a lot and to be honest, it will take around 2 years for you to get through all these. It took me 2 years and I’m basically using the same estimate. But trust me, AI is going to be the most consequential fields of this century and there is no reason for anyone to not get into this field.

There are two tips from my side when it comes to online education. Take notes and form study groups along with your peers. Yeah I mean it. Study groups work and nothing beats the power of collaborating while learning. Understanding different perspectives about each concept can transform the way you approach various problems.

Happy Learning!