Deep learning

UiT Machine Learning Group is active in deep learning research

figures-figure3

Our long-term aim is to advance basic research in deep machine learning and to develop new algorithms with a special focus on self-organizing deep networks in an unsupervised sense.

We are grateful to NVIDIA for GPU-donation to our group for deep learning research.

Our deep learning team primarily consists of

  • Michael Kampffmeyer
  • Filippo Bianchi
  • Sigurd Løkse
  • Robert Jenssen

where Michael has a primary responsibility and expertice, e.g. participating in the 2016 Deep Learning Summer School in Montreal. We implement our developed networks to a large degree using tools such as Theano and Caffe.

Example of recent paper

Semantic Segmentation of Small Objects and Modeling of Uncertainty in Urban Remote Sensing Images Using Deep Convolutional Neural Networks

Michael Kampffmeyer, Arnt-Børre Salberg, Robert Jenssen. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2016.

 

On the applied side, we are collaborating closely with the Norwegian Computing Center in Oslo. See also

Deep learning for applications

Comments are closed.