I went through whopping 5 hours of Stanford lectures on CNN for Visual Recognition [http://cs224d.stanford.edu/], as well as an hour of 'CS224d: Deep Learning for Natural Language Processing'. Both of these seem to have relevant content that I want to learn.
I should get on to doing proper hands-on solving soon though.
So far: I understand the generic procedure of Machine learning, a bit of backpropagation and normalization. Also the concept of neural networks.
I need to understand how the input data is chopped and diced, what features to extract as well as selecting a model for the training. And perhaps much more. Obviously.
Mood: Distracted, happy