Lecturer: Fred Hamprecht
Contents
Machine Learning is the foundation of certain developments that are amongst the most transformative for developed societies: think ubiquitious face recognition, surveillance, autonomous driving, etc.
Depending on prior experience and interest of the participants, this seminar will move between the following extremes:
1) An overview over pertinent machine learning methods: perceptron, support vector machine, decision and classification trees, ensembles, multi-layer perceptrons and deep neural networks.
2) An exploration of privacy-preserving machine learning (homomorphic encryption, differential privacy, etc.). The idea is to make informed decisions without directly revealing all measured features.
3) Budget-aware classification. Most research concentrates on maximizing accuracy. For embedded and handheld devices, and with a hunger for automated recognition in our billion devices, minimizing the computational cost / energy consumption becomes essential. This is becoming an increasingly important field of research.
Venue
Mathematikon Bauteil B, Berliner Str. 43, Raum B128
Time
Tuesdays, 11:15-12:30. We first meet on Tuesday, April 19th, 2016.
