LAVIS (Learning and Visual Systems)

Welcome to the Working Group LAVIS ("Learning and Visual Systems")

LAVIS' mission is smart data: Our expertise covers various aspects of data analysis pipelines, from signal processing over AI to user interaction. Our group is a member of the research focal area Smart Systems for Man and Technology and of the DCSM department at RheinMain University of Applied Sciences .

Prof. Dr. Ralf Dörner Prof. Dr. Ralf Dörner
Computer Graphics
Visualization
Homepage
Prof. Dr. Dirk Krechel Prof. Dr. Dirk Krechel
Content Analytics
Knowledge Management
Homepage
Prof. Dr. Ulrich Schwanecke Prof. Dr. Ulrich Schwanecke
Computer Vision
Mixed Reality
Homepage
Prof. Dr. Adrian Ulges Prof. Dr. Adrian Ulges
Machine Learning
Data Science
Homepage

Project Highlights

ATRIUM

ATRIUM addresses the real-time simulation and visualization of bulk materials following a new data-based approach based on machine learning instead of physically motivated equations. Using the results from classical calculations, training data is generated and a suitable regression model is developed to train a convolutional predictor. This promises a considerable speed-up, which e.g. facilitates more detailed and realistic visualization.

[more]

DeepWeather

Während numerische Wettermodelle eine mittelfristig gute Prognose großflächiger Trends erlauben, werden aktuelle Verfahren noch in sehr groben Rastern von mehreren km berechnet. In DeepWeather werden moderne KI-Verfahren auf Basis von lokalen Wetterdaten aus dem Internet-of-Things für eine hochaufgelöste Wettervorhersage entwickelt. Der innovative Lösungsansatz besteht dabei in der dynamischen Interpolation bestehender Wettermodelle mittels Regression durch künstliche neuronale Netze.

[more]

DeepCA

In Natural Language Processing, recent models based on neural networks (deep learning) open up possibilities for improving search, tagging and exploration. DeepCA's mission is to interlink neural models with conventional knowledge modelling using ontologies. A cognitive service for text and knowledge graph analysis will be developed that can easily adapted for the exploration of new domains and leads to innovative solutions for text annotation and search in heterogenous case bases.

[Website] | [more]