First Week of Udacity Deep Learning Nanodegree

On a whim I thought it would be a great idea…

I studied pure and applied math at Uni but never really got a chance to use it in the real world (went directly into more a CompSci job), but ever since I started reading some articles all the way back in 2004 I have had a keen interested in the AI/ML field.

Deep Learning has gone through many summers and winters to get where it is today but two key things happened to allow for this explosion in research firstly due to the overwhelming power of computers and the associated reduction in cost (thanks AWS) Deep Learning research and development can be done by anyone — this is really opening up the field to anyone with an internet connection and open mid.

Secondly — we are generating substantially more and higher quality data then we ever have, this means the amount of data available for researches and developers to train their neural networks with has grown hugely, of course a lot of that data isn’t publicly available (social networks, utilities, banks, governments hold huge untapped potential data sources).

In the past week the focus has been on getting the basic concepts and math right and getting our workstations setup (conda, tensorflow, numpy and a bunch of other dependencies)

Although the content has been good, I feel a few more examples really would have helped esp with the math of neural networks — thankfully plenty of content exists on the internet and I’m curating a github repo with links out to all the content I found useful. I also intent to publish my jupyter notebooks up on github with my learning on different areas.

I have some personal projects in mind I want to toy around with but I’m keen to move on to the Self Driving program after this (although I intend to build a mini self-driving car in between just to toy around with).