In collaboration with one of the leading meteorological companies in the world we are building a revolutionary new weather prediction system based on end-to-end deep learning. The approach is radically different from existing weather models and promises much higher prediction accuracy at a much lower computational cost. The practical test case is to significantly improve wind park energy production prediction. Read more
One of Peltarion's first major projects was a neural network based real estate prediction system. Released in 2009, it changed the industry and today in use by a majority of the Swedish banks, it featured an early version of what is today called "deep learning".
Seven years later we are proud to present the next generation that takes real estate evaluation to a new revolutionary level. Read more
In collaboration with the Zoological Department of Stockholm University we have developed a system that automatically detects and tags animals in photos taken by automatic camera traps. This is part of an conservation effort to track and understand the dwindling population of Scandinavian arctic foxes. The automated tagging replaces tedious manual work and increases accuracy. Read more
Experiments include generating choreography as using deep dream generated content as a medium that an artist can use. Read more