Machine learning is a broad topic, applicable to all data fulfilling some constraints. The power/possibilities are not just known after last year GO-challenge (google’s AlphaGo was winning) and DeepFace (facial recognition system created by Facebook). After this teaching unit, the participants should be able to see if something is a problem for machine learning or not, given a particular question and data. There will be a general cookbook and a deeper focus on tree-learners and SVMs. (There will be exercises for these two method classes.) The aim is that everyone is able to solve such problems, at least if it is a kind of plain problem for tree-learners and SVMs.

ML course content:

• the cookbook - about constraints, problem definitions, strengths and weaknesses of the main algorithm classes
• tree-learners
• SVMs
• exercises

About the speaker

Alexander Platzer
Medical University of Vienna
Austria
https://scholar.google.at/citations?user=lAfHAxoAAAAJ&hl=de&oi=ao

Dr. Alexander Platzer is currently building up a bioinformatics group at the Department of Medicine III at the Medical University of Vienna. He is never scared of Big Data, which is also indicated by his dissertation not so long ago (‘Analysis of most complete biological datasets’) and the large data collection of the 1001 Genomes Project (the final papers came out last year). He worked before at ftw (Forschungszentrum Telekommunikation Wien/Center for Information and Communication Technologies), which means he is experienced with several data sources of quite different types. Besides the different data domains (biology, medicine, physics, cryptography, video, network / telecommunication) the main questions remained the same: where does the data come from? Whereto does it go? And especially, why?