So here goes my first post!
When I first heard about deep learning and Google’s AlphaGo, I was skeptical about the capabilities of artificial intelligence. The idea that computers could learn without being explicitly programmed seemed too unreal to be true. This was especially the case given that I was taking a computer science course in high school, where students were expected to write if-else statements and the variants. If a human programmer doesn’t instruct the computer to perform a for-loop ten times, how would a computer “know” to perform this operation in the first place?
This question has nagged me for a very long time. It has nagged be quite a bit, actually, to the point that I have decided to study the inner workings of deep learning on my own.
Self-studying Deep Learning
This little project, inspired by Daniel Bourke’s post, “My Self-Created Artificial Intelligence Masters Degree,” encapsulates my journey towards uncovering a mystery that is deep learning. I would not call my spare time self-studying to be equivalent to working towards a masters degree, but the idea of developing a curriculum and outlining a clear set of goals and timelines seems like a great way for self-motivation.
What, then, would I be studying in particular?
Based on cursory research and my current level of superficial understanding, deep learning might be described as a burgeoning interdisciplinary field that is a marriage of computer science, statistics, linguistics, and infinitely many more. At its core, however, deep learning algorithms are undeniably buttressed by the field of mathematics, specifically linear algebra, calculus, and probabilistic theory. Built on top of these theoretical foundations, deep learning utilizes the power of modern computing to process big data and identify patterns that help make accurate predictions.
Of course, the picture that I have just presented is by no means extensive, yet one point remains crystal clear: as a student with very shallow programming background (a secret: I barely got this website up and running on GitHub pages) and knowledge of just introductory college-level math, self-studying deep learning is going to be a challenge, and a fun one indeed.
Which naturally leads to the question: from where do I begin?
The First Book
The first book to be added to my curriculum is The Hundred Page Machine Learning Book, written by Andriy Burkov. As the title suggests, this book provides a highly condensed guide on the clockwork of machine learning. Whether the contents of this book will be a significant challenge given my current level of knowledge is yet unclear, yet it will surely show me an overall glimpse of what studying machine learning will entail for me.
There are other books and online resources that I would like to study, but introducing a comprehensive list of such resources, alongside a justification behind each selection, will be tabled for separate post. Come back in a few days for a hot, freshly-baked book review of one of the best sellers on the art of machine learning!