I came across this article:https://research.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html, and was intrigued by the machines 'idea' version of what it was trying to recognise. This has inspired by Extended project qualification (http://www.aqa.org.uk/programmes/aqa-baccalaureate/extended-project/the-aqa-epq). I shall be completeing Kaggle's dog vs cats recognition test and comparing each model's effectivness with their visulisation.
Since being unable to compile Caffe I will use a pre made docker container as my starting point to overcome this. the orignial container comes from herval (https://github.com/herval/deepdream-docker).
Have also used code from adilmoujahid's tutorial (https://github.com/adilmoujahid/deeplearning-cats-dogs-tutorial)