IC Colloquium : Machine Learning Approaches for Security Problems
Event details
| Date | 14.03.2013 |
| Hour | 16:15 › 17:30 |
| Speaker |
Mario Frank, University of California Berkeley IC faculty candidate |
| Location | |
| Category | Conferences - Seminars |
Abstract
For many security problems it is increasingly difficult to manually examine the ever increasing amount of relevant data. At the same time, noise in the data disqualifies deterministic automation approaches. As a consequence, the development of algorithms and systems that can learn and generalize from a large amount of noisy data constitutes one of the fastest growing fields of security.
In my talk I will outline this field and argue that it is an integral part of security. I will give examples of security problems where, already today, the best solutions involve
unsupervised or supervised learning methods. My examples reach from local touch-based authentication for smart phones over mid-scale policy learning for enterprise systems to large-scale detection of malicious accounts in online social networks. My talk highlights open problems that will guide my future research agenda.
Biography
Mario Frank is currently a post-doctoral researcher in Dawn Song's group at the University of California, Berkeley. Before joining UC Berkeley in 2011, he was a PhD student at ETH Zurich. His thesis ``Probabilistic Role Mining'' was awarded the ETH Medal. Mario studied physics at the Ruprecht Karl University of Heidelberg and
the University of Sydney. In 2007, he received the Otto-Haxel Preis for graduating best-of-class and for his diploma thesis about a 3D camera.
His research interests lie in the intersection of security and machine learning.
For many security problems it is increasingly difficult to manually examine the ever increasing amount of relevant data. At the same time, noise in the data disqualifies deterministic automation approaches. As a consequence, the development of algorithms and systems that can learn and generalize from a large amount of noisy data constitutes one of the fastest growing fields of security.
In my talk I will outline this field and argue that it is an integral part of security. I will give examples of security problems where, already today, the best solutions involve
unsupervised or supervised learning methods. My examples reach from local touch-based authentication for smart phones over mid-scale policy learning for enterprise systems to large-scale detection of malicious accounts in online social networks. My talk highlights open problems that will guide my future research agenda.
Biography
Mario Frank is currently a post-doctoral researcher in Dawn Song's group at the University of California, Berkeley. Before joining UC Berkeley in 2011, he was a PhD student at ETH Zurich. His thesis ``Probabilistic Role Mining'' was awarded the ETH Medal. Mario studied physics at the Ruprecht Karl University of Heidelberg and
the University of Sydney. In 2007, he received the Otto-Haxel Preis for graduating best-of-class and for his diploma thesis about a 3D camera.
His research interests lie in the intersection of security and machine learning.
Links
Practical information
- Informed public
- Free
- This event is internal
Contact
- Christine Moscioni