I wrote this book because on my long and winding deep learning journey I learned some things that I believe can be useful to others. Many of the insights that I share can accelerate your progress several-fold.

In 2012, I couldn’t program. I had a corporate job that provided a steady stream of income, but it was utterly boring. There had to be more to life than thinking of human beings as resources. I read somewhere about these things called MOOCs, and I decided to take a look.

By 2018, I had done enough deep learning to win a Kaggle competition,[1] just a few months after a blog post I wrote while taking a course got published by KDnuggets.[2]

In 2019, I received an invitation to visit my colleagues at Curai, a healthcare startup in San Francisco. Out of around 20 engineers, I was their only team member from overseas.

flying to US
Figure 1. Flying to visit my colleagues in the US

As I write these words, I have the most amazing job I could imagine. I get to work on decoding non-human communication as part of the Earth Species Project with some of the most amazing, interesting people I have ever met.

The reason why I believe I have something worthwhile to share with you is not, however, related to the aforementioned achievements.

Over the years, I’ve learned a lot about learning and acting effectively. This change – and the ideas that led to it – are the essence of what I would like to tell you about.

At the beginning of my journey, I moved very slowly. I studied for months, only to discover that my deep learning abilities had not progressed much. I couldn’t chart a path that would take me from where I was to where I wanted to be. Then, on one of the many detours I took, I became a web developer because I had started to believe machine learning might be out of reach for me.

What allowed me to persevere was my belief in learning. Even if I wasn’t learning quickly, I was still learning, and that was something.

I started to reflect more and more on what worked for me and what didn’t. I also managed to keep an open mind, always ready to accept advice (though that might have been more out of desperation than anything else). Often, I would write my thoughts down and share them with others.

tweet kjbird15
Figure 2. A tweet from a member of the community. Thank you, Kevin!

This process of introspection was key in motivating me to continue. Through it, I began to employ new ideas, and refine my approach.

I had to cover a lot of distance, as I don’t have a college degree. I only started to learn to code when my first daughter was about to be born. This gave me the focus I had lacked earlier, but it also placed some rather significant constraints on my time.

If my approach to learning hadn’t undergone such an immense change, I would not be writing these words to you today.

In this book, I share the life-changing ideas I picked up along the way. They will allow you to learn faster and to accomplish more in a shorter period of time. They can alter the trajectory of your life, just as they have mine.

How to read this book

"You can’t connect the dots looking forward. You can only connect them looking backwards"

The chapters of this book are loosely related, and can be read in any order. The best approach would probably be to read them in sequence, then to return and review chapters of interest over time.

To help the ideas I describe permeate your life, talk about them with your friends or loved ones. Tweet about a thought with which you agree or disagree. Maybe write a blog post elaborating on one of the concepts. I continue to experience firsthand how powerful the processes of introspection and sharing with others can be.

Change is gradual, and will not happen overnight. Pick an area you would like to work on, give it some time, and, only after a while, move to another.

Above all, be gentle with yourself. Most of the time, I feel like I’m standing still. When you view your life from the perspective of days or weeks, it can be really hard to notice progress. But, you need to start doing just a few things differently on a regular basis to start getting exponentially different results.

There is no superhuman effort required. Do a few things differently, and, very soon, you will be able to look back and start connecting the dots that brought you to where you ended up.

1. The iMaterialist Challenge (Fashion) at FGVC5 was a multi-label, fine-grained computer vision problem. The competition was fierce, but I managed to come in first place by a wide margin. You can read about my solution here.
2. How to do machine learning efficiently - this article describes how to structure your machine learning project to achieve the best results in the least amount of time.