Hands on Deep Learning

12 Jun 2015 . Machine Learning . Comments

It has been over a month now that I have this Youtube tab open (among at least 100 others) which is about Deep Learning Lecture at Oxford by Nando de Freitas. This was the time I was alone in my room on sick leave for over a week, I decided it was time to do some good stuff. Although, I am not very impressed by the lectures so far, I was informed that all the new Deep Learning stuff is done in Torch7. And thats how I started learning this amazing tool.

The first steps as conveniently listed on the Cheatsheet included learning lua. lua is a lightweight multi-paradigm programming language designed as a scripting language with extensible semantics as a primary goal. Lua is cross-platform since it is written in ANSI C, and has a relatively simple C API. I figured it was worth sometime spending time learning lua and getting familiar with the language. I have written another post Becoming familiar with lua in which I list the web resources and a fun implementation of a command line game.

Update 22-05-16

I have since then come a long way. Implemented lots and lots of Deep Learning Architecture in Torch7. This included mastering nn, nngraph and other Torch7 libraries. I have made a contribution to nngraph. This happened as a gradual process while working on Deep Conversational Models for my company TrueAI. In future, I might opensource some of the generic modules I wrote.