Work on powerpoint with introduction

[dafnevk]

Oral explanation in talk or during workshop: Explain what deep learning tries to solve why should researchers be interested in learning / using deep learning they need classification what kind / group of research questions fit why not use classical statistics (classical vs DL approach) To avoid creating customized features, if desired & To create novel feature you may not have thought about & To classify complex pattern constructed from simpler patterns (and give cat example) DL is also partly unexplored terrain and therefore may add novelty to their field Explain what mcfly tries to solve Ease the hyperparameter and architecture choice Why would I use it instead of caffe tensor flow Specific for time series It uses tensorflow and keras, and mcfly provides an easy access way to start learning Keras (it has a less steep learning curve) Up to date with latest trends in 2016 informed by literature use expertise to make the choices so that the users don’t have to Explain Logistic regression and/or single node in a neural network?

[vincentvanhees]

Issue moved to NLeSC/mcfly-tutorial #3 via ZenHub