Data Science Week 2019 – „Deep Learning”

This year's Data Science Week is dedicated to the topic "Deep Learning" and will take place for the second time from 9-11 October in the Hessian Rhön. The focus will be on the question: Can microorganisms be detected with the help of neural networks and how can this contribute to measuring the water quality of water bodies?

Neural networks are increasingly used to analyze large amounts of data and solve complex practical problems. For this purpose, increasingly complex and computationally intensive neural network architectures are used – deep learning is the key word. The higher requirements come in particular from the areas of automated image processing, which serve to identify objects, patterns or even people and 3D environments. The perspectives are immense: objects can be identified, marked (tagging) and tracked – up to the prediction of their further behavior.

In this 3-day event, we will focus on the analysis of microorganisms in bodies of water. Specially developed deep learning nets will be used to track down paramecium and the like. Light microscopic images of water samples are taken and analyzed with neural networks. The organisms in the samples are to be identified and typified.

Technologically, this year's Analytics Event is dedicated to the topic of image processing with neural networks and raises questions about the theory of neural networks:

• Why deep learning when neural networks are universal approximators?
• Which network structures best achieve a high model quality?
• How dependent are the models on the field of application - how strong is the influence of overfitting?

as well as the practical application in the domain of microorganisms:

• How meaningful are the results?
• How long and with how many samples must a net be trained in order to be able to make reliable statements and how can these be collected efficiently?
• Can water quality be detected better or faster than with conservative methods?
• What economic uses are there?

As a result, a micro service prototype will be developed to analyze images of microorganisms through the deep learning network we have developed and to continuously optimize the model quality of the network.

Back to news overview