Machine translation, image recognition, product recommendations ... more often than not there's deep learning involved.
As a consequence, demand to integrate deep learning technology into "normal" applications will grow. Not many software architects, however, have experience with all major frameworks. Here, it helps to know the concepts involved. How does it really work?
This is what the session is about: background knowledge, concepts, essential mechanics. And based on that understanding, a high-level overview of the frameworks.
Target Audience: Architects, Developers, Project Leaders, Managers, Decision Makers (everybody!)
Prerequisites: Curiosity about the topic; not being scared off by a little math
If you're asking, what's the hype in 2017? - chances are somebody will say "deep learning".
Image recognition, machine translation, strategy games like Go - all these have recently been taken over by deep learning.
Automated music generation, colorization of black-and-white films, product recommendations - more often than not there's deep learning behind.
As a consequence, demand to integrate deep learning technology into "normal" applications will grow. But what is deep learning technology?
There's DL4J for Java, Torch for Lua, TensorFlow, PyTorch, Caffee, Keras (and more) for Python ... too many for the software architect to know them all.
Is that a problem? Probably not, as like in many cases the actual choice of framework will depend on factors like the surrounding architecture, performance requirements, developer skills etc.
What's special with deep learning however, is the subject matter involved - what even is deep learning? How does it work? What do I have to do to make it work?
This is what the session is about: Provide the background knowledge, explain the concepts, explain the mechanics. And based on the concepts, give a high-level overview of the frameworks: How do they differ - in prioritization, implementation, developer friendliness?