Deep Learning @Mobiliar // Everything is data, everything is a graph

Session 1:
Deep Learning @Mobiliar //
Speaker: Paolo Kreth, Schweizerische Mobiliar//
Session overview:
At Mobiliar the number of use cases that deal with AI are rapidly increasing. We plan to solve them using a machine learning approach. One branch of machine learning ist the so called "Deep Learning", which relies on the usage of neuronal networks. Deep Learning often finds application in fields like text analysis and image recognition. During the calculation of a neuronal model we have to deal with very intensive vector transformation. Those operations can be accelerated by factors through the usage of special hardware, so called (Graphical GPUs). Normally a company does not have a GPU infrastructure on premise. Therefore it is the perfect user case for a cloud approach. During a semester work for a CAS in Big Data different cloud infrastructures for deep learning were tested and evaluated. We will tell how the evaluation criterias where chosen, and show the result of the evaluation.
//

Session 2:
Everything is data, everything is a graph //
Speaker: Sabrina Senna, Oracle Labs //
Session overview:
Stability is no longer the benchmark of success in business. Nowadays, a business has to focus on innovation and rapid development if it wants to obtain success. But how do companies adapt and keep up with technology change? The answer is Big Data. Exploring data can give an enterprise multiple insights giving competitive advantage in the industry. But how to analyse this data? As Graphs. Graphs are the most powerful form of displaying exploring and analyzing data, allowing information aggregation over heterogeneous data sources, powerful visualizations and interactive exploration. Based on the importance of graphs in Data Science, Oracle Labs developed multiple technologies that enable Data Scientists to accelerate the process of exploring and gaining insights from data. These technologies are PGX - Parallel Graph AnalitiX, a graph analytics framework and Oracle Labs Data Studio, a web-based notebook platform that enables live code collaboration in multiple programming languages with graph analytics and rich, interactive visualizations.
//