Benedek Rozemberczki: The Peril and Promise of Machine Learning
In the last few years we observed a revolution regarding machine learning algorithms that use social network data. Emergent methods can do inference on users who share very limited data about themselves, yet achieve high performance on predictive tasks. As part of the presentation we will take a whirlwind tour on deep learning procedures that use graph structured data. Using publicly available data from music streaming services and a swinger dating site I will demonstrate the immense power of these models with real life use cases.
Followed by a panel discussion involving:
Balazs Lengyel (Institute of Economics, Hungarian Academy of Sciences)
Chrys Margaritidis (Dean of Students at CEU, Research interest: ethics of big data)
Matteo Neri (PhD student, CNS)