Deep Learning

Deep Learning (DL) is a sub-field of Machine Learning. Deep Learning uses neural networks to analyse large data sets. The structure of the artificial neural networks is inspired by biological neural networks. Thus, by using neural networks, human behaviour, such as reasoning from information, can be imitated.

A neural network has the advantage that much larger mountains of data can be examined much faster than humans ever could. Thus, neural networks are used wherever patterns or trends in large amounts of data are examined, for example in face, object or speech recognition. During the use of a neural network, it learns from the input data and improves its performance (self-learning system).

Symbolic representation of a neural network

Try it out for yourself!

https://playground.tensorflow.org

Regional experts

Prof. Dr.-Ing. Jürgen te Vrugt
FH Münster
Teaching and Research Area Artificial Intelligence
Contact

Prof. Dr. Xiaoyi Jiang
WWU Münster
Dean of the Department of Mathematics and Computer Science
Contact

Applications

Forest AI - object recognition
Kaitos GmbH has developed a forest AI that uses drone images to detect damaged or diseased trees in a forest and provides them with a GPS tag.

Automated quality inspection
By using neural networks and computer vision, deviations from the target state can be determined camera-based. This can be done very easily through Amazon Lookout for Vision. All that needs to be done is to provide two folders of images. One folder with good parts and one folder with bad parts. Then the pre-trained neural network is adjusted by the input of the images. In this way, deviations can be detected.