How I Quit my Law Studies to Learn Online what I Love: AI

And why you should do the same.

There are 2 parts in this article :

  • My story
  • All my strategies, advices and resources for self-taught. If you want to jump directly on this part click here

Your “Geek Stuff”

“You want to be a programmer? Really? But you’re bad at math, you don’t have a brain to do math, don’t expect to work in this domain, do something else.”

I heard this phrase when I was at my last year of high school, during an appointment with an orientation counselor. I told him that I love to program and I learned to do it in C since my 13th with my good old Windows 98’s computer and Visual C++ for Windows 3.1.

Good Old Days
“That’s cute… But maybe is better that you do something where you can have a chance to be accepted and have a diploma. You can do your … well… ‘geek’s stuff’ as a side project”

Indeed, at the end of high school, like all students, I had to choose a course of studies for University. I had a very good record but I couldn’t even try engineer studies because my major was economics and history, not mathematics.

Consequently, after one uncomplete year of Bachelor’s degree of Geography, I took a Bachelor degree of Law and Political Sciences. Why? Well…

“You know law is the perfect studies for people like you who don’t know what to do because you can be a lawyer, a jurist, a judge… It’s a very interesting topic. Moreover, you have a literature brain consequently it will be really simple for you”

Simple? Not at all. Interesting? Neither. I hated these studies, law was for me a very hard and really not interesting topic such as political studies.

Do you know Turing ?

At the end of my third year, I learn about Alan Turing and his Turing Test in a CNRS French Documentary. I was completely fascinated about his “Turing Test” paper.

The article on the Book: Computers and Thought

I do a lot of research about AI and at this time I found this topic fascinating: all the revolutions that are going to happen in all fields: the self-driving cars, in education, in health, in research, in space exploration… The GANs, the music generation… I want to be part of that.

Yes, but at this time, I thought that I can’t do that. I must face it, I’m a student in Law school, ok I program stuff, but I’m bad at maths how could I dare to try to learn AI? That’s stupid…

You should watch this course: “Learning How to Learn” it’s very interesting

One of my friends told me about a course “Learning How to Learn” by Barbara Oakley and Terrence Sejnowski.

This course and her book was a breakthrough in my life. It was the first time that someone says to me that you can learn everything you always wanted to learn. It’s just a question of passion , hard work and good learning methodologies.

The biography of Barbra Oakley was very inspirational, she hated math (like me) and she began to learn it at 26 years old and became very good at it.

By using scientific proof, she demonstrated that there is no such thing like Math brains people // literature brains people.

I watched her course in 3 days, I couldn’t stop until the end and at this time I understood that it was time : I must take a gap year for doing what I want to do: AI.

I pass my exams and at the end of my Bachelor’s degree I had two choices:

  1. Continue to studies and a career that I will hate all my life (Law).

2. Do what I want to do with a gigantic risk of failure (AI studies).

I took the second option

Mom, Dad, I quit University and I will learn Deep Learning and AI with Udacity

“Uda Quoi ?! (Uda what ?!)”, My Dad
A realistic representation of my dad’s reaction

It was one of the hardest thing to do: saying at my parents that I quit college and begin to learn what I love.

Why? Because parents in general are people who are extremely worried about the future of their children and telling them that you quit a path that leads you to a stable and high paid job to an online formation in a mysterious field is a natural source of anxiety 😄.

But one of my biggest chance is having the greatest parents of the world. They always let me trying things. And they gave me one year.

5,4,3,2,1 Let’s go (#5SecondRule Mel Robbins)

“This is the moment, this is it” Michael Jackson

September to December : Math math math

Gradient Descent

This was the first step, I was so bad at maths, sure I know how to calculate, but seeing any formula was a nightmare for me.

I learn algebra with Khan Academy, Calculus and Probabilities with MIT OpenCourseWare. And surprisingly I begin to like it.

Well… Maybe not as much as her 😅

January- May : Deep Learning Foundations Nanodegree

Exemple of projects : Image classifier / Face Generation

In January, I began this my formation in Deep Learning (Deep Learning is a fascinating part of AI) with Udacity.

This Nanodegree was a 5 months intensive course composed of acquiring the basic and advanced topics of Deep Learning theoretically (mathematics abstractions) and practically (a lot of projects to do mostly with TensorFlow).

  • Basic Neural Networks (NN)
  • Recursive Neural Networks (RNN and LSTM)
  • Sequence-to-sequence
  • Convolutional Neural Networks (CNN)
  • Generative Adversarial Networks (GANs)

5 projects:

  • First Neural Network (NN)
  • Image classification (CNN)
  • The Simpson’s script generator (RNN)
  • English/ French translator (Seq to Seq)
  • Face generator (GANs)

This intensive Nanodegree was so interesting and so good! The projects were hard but really fun. If you want to learn Deep Learning my advice is take this Nanodegree (Disclamer: I have NO sponsor relationship with Udacity).

May — : AI Nanodegree and what next?

I’ve just started my AI Nanodegree it will be a 2 term formation.

So what next? Many things:

  • An open source project: The Simpsons Synopsis Generator 📺, an AI that generate synopsis for the Simpsons.
Wait what ?!

Well… As you can see my AI generates very strange synopsis 😨. It needs more improvements.

  • I’m starting to search for job opportunities in the US and all the administrative consequences that this entails.

Doing what you love is such a release, it is an immense pleasure but don’t forget that it is also hard work. I work 12 hours per day, but I’ve never felt so happy, fulfilled and motivated as I am today.

My advices and resources to self-taught

You have no excuses!

First of all, you have no excuses : you can learn everything with internet: from the Uzbek language to the study of nuclear fission. From MOOCS to tutorials on youtube you have access to a ton of free resources.

You want to create a video game? You can learn it here! You want to speak French? You can learn here.

Plan well you’re not in vacation!

“By failing to prepare, you are preparing to fail.” 
Benjamin Franklin

Plan well and stick to your schedules you’re not on vacation!

It’s maybe the hardest thing : because there is a ton of resources, it’s simple to be overwhelmed.

My methodology:

Have 3 plannings

  • One for the month: what you want to accomplish in each domain ?
A glimpse of my Month todos
  • One for the week, be honest and realist about what you can do.
  • A Todolist for each day: done the day before, each task must be estimated in time.

As you can see, you don’t need fancy programs: I use excel for my Monthly and weekly planning and Notepad for my todolist.

Work Hard Play Hard

Your time is precious. You must be fully commited on what you do if you want to be productive and get the most of the day, there is nothing worse than spending hours on learning a topic but remembering nothing because you were not focusing on it.

Word Hard Play Hard

When you’re working, work hard, be focus, turn off your facebook, phone, snapchat and all of that stuff that destroy your focus. Be fully commited on what you do.

When you’re playing, play hard, have funny distractions that you really like. No spending time on facebook is not a funny distraction! You know it, it does not bring much pleasure.

You like the Sims, Starcraft? Better idea to play 10 minutes the Sims than spending 10 minutes on Facebook.

Use the 50/10 Rule

It’s scientifically proven that your brain can be concentrated in maximum 50 minutes. It’s really useless to work hard during 5 hours without breaks.

My advice use the 50/10 Rule:

50 minutes of work

10 minutes of pause

For God sake culture is not a fancy word!

Cultivate yourself! The worst thing that can happen to you is to become a specialist in your domain but an uneducated in all other fields.

The best people are whose who are cultivated in different fields which allows them to see the world in different perspectives and as a consequence, to innovate.

Listen good podcasts like the NPR’ podcasts, go to museums, go to the library, READ READ READ!

Have a social life

Sure you’re busy, but your friends and family are the most important supporters that you have. So take care of them and take care of yourself. It’s not productive to study all days inside your flat.

Moreover, if like me, you transit from a fully social life in University to an online training working at home it can be really hard so you need to go out, to see people.

Even if I’m really busy I try to go out every week to drink with friends, to go to cinema etc.

Do Side Projects !

Side projects are great especially in the programming field. It helps you to apply what you’ve learn. But also hoping to raise your profile. And sometimes making money.

2 importants things to consider:

  • Start small then improve: sure you have this great idea of creating the next facebook with a ton of features. But start small or you’ll never finish.
  • One side project at a time!: sure you want to create a video game, but also write a book, but also create a new AI agent. That’s cool, that’s great idea, but tell me how you’re gonna to handle this 3 projects ? Pick one and finish it!

Resources

MOOCs:

Learning better, learning faster:

  • Learning how to learn : one of the greatest MOOC of all times. Barbra Oakley give you a lot of keys to learn better and faster.
  • Mindshift : the continuation of the first MOOC.
  • Feynman Technique : this is for me the best technique to learn everything.

Learning mathematics:

  • Khan Academy, is a free MOOC platform with tons of videos about a lot of topics
  • MIT OpenCourseWare is the MIT platform with a lot of great courses, Calculus and Multivariate Calculus are the best.

Learning Deep Learning/AI:

That’s all, I hope that this article inspired you and give you some useful advices to be a self-taught. Remember that doing what you love depends ONLY on your choices so affirm yourself and do what you love ! 😄

If you have some experiences you want to share, feel free to comment this article I want to hear from your experiences 😄.

And If you liked my article, please click the 💚 below so other people will see this here on Medium.