1. Who is better ? Human or computer ?
This article is the first among a series of articles, through which I believe I could share what I know in deep learning in a very simplified and casual way, so that no one else should sweat to grasp the fundamental ideas of deep learning as I did. There are already a few other people who does the same simplification proces of these concepts, and I would like to contribute this series as well, to that collection. I will be focusing on these two things while writing this, 1. not to exclude underlying math entirely ( including simple explanations ) 2. demo coding in simple languages, lua (Torch) & processing.
A tiny intro :
I’ve started with a question. For me, the answer will be, “Well, it depends”.. Here’s an extension to it: computers are very good at finding results like square root of a number, complex divisions, which a human can’t do that easy. Whereas, we’re far good at recognizing patterns ( like finding a face in a big crowd even if the lighing is unusual ) and making appropriate decisions. Computers are perfect to follow a rule whereas we’re good at approximating.
So, what fundamentally keeps human and machines on two different sides?
Hint: We've got a brain which works differently.
( Now I think of a related question: Why do we want machines to be more human like ? by making the mistakes we make and compromise the perfection ? )
Narrowing down to deep learning
Nowadays, with the advant of cheap hardware including GPUs, deep learning is a hot and state of the art topic among machine learning community, it proved its dominance over many of the standard ways of solving much complex problems, and this happened very recently. Deep learning is what we’re interested in, which makes machine behave like human experts for a given task. The researchers were able to draw cartoon of brain called Artificial Neural Networks ( although, we’re still trying to understand how actualy our brain works ), and thats what we are ging to learn. Buckle up !
–
This series of lectures were written as a transcript of my talk during #planC 2016 @ MCA CET.
–