23rd Deep Learning Meetup in Vienna

Dear Deep Learners,

We start the new year with two exciting topics: Explainable Deep Learning and an extensive report from the number one AI conference, the NeurIPS:

Talk 1:
Explainable Neural Symbolic Learning
by Ahmad Haj Mosa (AI Researcher, PwC Austria), Fabian Schneider (PoC-Engineer & Researcher, PwC Austria)

Explainable Deep Learning is getting more and more important. We show that the explainability vs. performance of a model is not an orthogonal but a beneficial relationship. Additionally we propose a neural-symbolic framework for implementing this relationship by using "Object Oriented Learning" and Haskell.

Talk 2:
Report from NeurIPS conference
by Rene Donner (Head of Machine Learning & Engineering, Contextflow)

Reporting from the recent NeurIPS conference (formerly NIPS, largest AI conference in the world), we will present new trends and methods, including advances with are easily applicable in your current projects.

Topics include:

• Interpreting neural networks as differential equations
• New regularization techniques
• DL based image registration
• Distributed training
• Video to Video GANs
• Image generation from text

After the talks there will be time for networking and discussions. We thank Thomas Faast, Innovationsmanagement Fachhochschule Technikum Wien, for hosting us and providing drinks and snacks.

If you have hot topics to present or announcements to make please let us know beforehand.

Looking forward to kick-off the 2019 Deep Learning Meetup season,
Tom, Alex, Rene, Jan