Data Science Meetup Hamburg

=== Doors open ===

=== Small intro ===

=== Talk 1 ===

Miro Zoričák on N4 Chatbot

Recent improvements in the area of Natural Language Processing gave rise to the new generation of Chatbots. We will look at how we have approached the problem of building a Chatbot for our customers here at enfore using rule-based systems, Word Embeddings and Deep Learning. Which approaches worked, which ones didn't and why.

=== Talk 2 ===

Stefan Kühn on "Deep Learning and Optimization Methods"

It is a common belief that SGD - Stochastic Gradient Descent - and related improved methods like Momentum, Adagrad or Adam are the best choices for training deep and potentially complex networks. In this talk we want to show that the Optimization Toolbox has more to offer, and that advanced methods like Limited Memory BFGS (L-BFGS) and Conjugate Gradient (CG) can indeed speed up training times in many cases if applied properly. We do not do a deep dive into theory but focus on practical aspects. And don't be afraid of the additional complexity - you have already implemented a deep learning method!

=== Networking ===