Gæsteforelæsning: Mikkel H. Brynildsen, Chief Data Scientist, Big data & AI Lab, Grundfos
05.04.2019 kl. 14.30 - 15.30
Intelligent Water Solutions: Data & Knowledge Driven AI at Big Data Lab, Grundfos.
Grundfos develops water solutions for the world. Grundfos is a pump company connecting with millions and millions of people every day.
I will give examples of the challenges Big Data & AI Lab in Grundfos is working with. This can be in digital products: predictive maintenance of pumps and water systems, energy saving by optimizing pumps systems, detecting and localizing leakages - but also in running the business smarter: communicating more efficiently with customers, or making our many years of experience available on the semantic web. At Big Data & AI Lab we work mainly with machine learning (ML). Both on learning coefficients in physical models (regression) and more purely data driven systems (decision trees, convolutional neural networks, reinforcement learning, ...). Besides ML, Big Data & AI Lab also works with what is known as "Symbolic AI". "Symbolic AI" is the art of manipulating knowledge bits (symbols) like facts, rules, concepts, logic, and probabilities in order to gain more knowledge, making a system work "intelligently". An example is expert-systems that implements business-rules when run on input data. Historically, symbolic AI was among the first efforts to emulate how humans think. Symbolic AI is not optimization based (in general), and therefore not as data intensive as machine learning.
We are starting to see the need for a combination of these kinds of AI (ML and Symbolic AI) in hybrid systems; for instance using OCR and CNN to extract information from PDFs, then using rules to act on the insights in a second layer of a hybrid system.
I hope the talk will open the floor for some fruitful discussions about how Big Data Lab and AAU can raise the bar in Intelligent water systems together.
Institut for Matematiske Fag
Skjernvej 4A, lokale 5.018