Demystifying Deep Neural Networks – Rosie Campbell / Front-Trends 2017

Deep neural networks are the latest hot computing trend – but to most people they are still a bit of a mystery.

With the release of libraries like Google’s Tensorflow and the open source ConvNetJS, there’s never been a better time for developers to start exploring deep learning. But I know from experience that wading through complex academic explanations can be pretty intimidating!

That's why I wrote a talk that starts from scratch and assumes no prior machine learning knowledge – but by the end of it, I aim for you to go away as excited about neural networks as I am, and knowledgeable enough to rave to your friends about them in the pub.

We'll look at what makes a neuron, how neurons become a network, how networks are trained and what the 'Conv' in 'ConvNets' means. We'll see how ConvNets can produce weird and wonderful works of art, how they can turn your selfies into masterpieces, and look at some of the rather amusing ways they can go wrong!