Sicco Verwer Home

I moved to TU Delft, my new homepage is here:
http://cys.ewi.tudelft.nl/users/sicco-verwer

I am a postdoctoral researcher at Radboud University Nijmegen in the Department of Model-Based System Development (MBSD) at the Institute for Computing and Information Sciences within the faculty of Science

My expertise is in grammatical inference, machine learning, data mining, and artificial intelligence. Specifically, I am interested in learning complex state machines (including timing and parameters), using learning for more than just prediction (for instance model checking), and the power of search methods (using SAT-solvers and Mixed Integer Programming) in machine learning.

NEWS

I got awarded a VENI grant from STW for my project on Learning State Machines for Network Traffic Analysis (MANTA)!
Together with the national cyber security centre at the department of security and justice in the Netherlands and Surfnet, I will investigate methods for learning timed state machines automatically from network traffic, and searching for malicious behavior in these machines using modern model checkers.

I help with the organization of the RERS reverse engineering of reactive systems Challenge. The competition provides an interesting setup that aims to combine state machine learning methods with model checkers and code analyzers!

My LEMMA project Learning Extended State Machines for Malware Analysis got awarded in the recent NWO cybersecurity call!
Together with experts from Madison-Gurkha, Thales, the department of security and justice (WODC and NCSC), and Surfnet, we will build tools for learning state machines from network data and combining the knowledge contained in them using information fusion techniques. The goals are to localize the presence of Malware within a network infrastructure and to pinpoint possible sources for these infections.

Our Careful project got awarded in the recent NWO open competition call!
We will investigate methods for automatically learning a followed medical protocol (essentially a state machine) from patient data from the NIVEL institute. Such models will be used to highlight differences between the envisioned and actual models of care.