BC3: Fun with Machine Learning
Machine learning (ML) aims for technologies which can model/improve the behaviour of algorithms based on examples. Apart from prominent paradigms such as deep learning, ML covers a zoo of different methods; these help to solve diverse tasks such as fault prediction, control of technical devices, data inspection, playing games, computer vision, or speech processing. The goal of the course is to give an intuition about principles of machine learning, and to exemplarily present a few algorithms and their applications in different areas of machine learning. The content of the four sessions will cover the following questions:
Session 1: Foundations of Machine Learning: What is machine learning? When can it be used? What are the pitfalls?
Session 2: Classical learning paradigms: a gentle introduction into the all times top ten of machine learning
Session 3: A spotlight on advanced approaches: deep learning and recursive networks
Session 4: The power of metric-based approaches
The aim of the course is (1) to raise awareness about when machine learning can be used and when it is better avoided, (2) to introduce to fundamental paradigms and their application domains and application pipelines, (3) to give an idea about current hot spots and open questions in machine learning.Literature
Christopher M. Bishop. 2006. Pattern Recognition and Machine Learning (Information Science and Statistics). Springer-Verlag New York, Inc., Secaucus, NJ, USA.
Patrick O. Glauner, Manxing Du, Victor Paraschiv, Andrey Boytsov, Isabel Lopez Andrade, Jorge Augusto Meira, Petko Valtchev, Radu State: The Top 10 Topics in Machine Learning Revisited: A Quantitative Meta-Study, https://arxiv.org/abs/1703.10121
Ian Goodfellow and Yoshua Bengio and Aaron Courville, Deep Learning, MIT Press 2016, http://www.deeplearningbook.org
M. Biehl, B. Hammer, T. Villmann, Prototype-based models in machine learning,
Advanced Review in WIRES Cognitive Science, 2016, doi: 10.1002/wcs.1378
V Losing, B Hammer, H Wersing, KNN Classifier with Self Adjusting Memory for Heterogeneous Concept Drift, IEEE International Conference on Data Mining (ICDM) 2016
Barbara Hammer heads the research group Machine Learning at the CITEC centre of excellence at Bielefeld University. She received her Ph.D. and venia legendi (permission to teach) in Computer Science in 1999 and 2004, respectively, from University of Osnabrück, before accepting an offer as professor at Clausthal University of Technology in 2004. Several research stays have taken her to Pisa, Padova, Groningen, Paris, Bangalore, and Birmingham. Her fields of interest cover recurrent and recursive networks for structures, self-organizing maps, data visualization, learning interpretable models, and incremental learning, clustering as well as applications in bioinformatics, industrial process monitoring, and cognitive science. She particularly enjoys research projects in an interdisciplinary team.Website