Prof. Emo Welzl and Prof. Bernd Gärtner
|Mittagsseminar Talk Information|
Date and Time: Tuesday, December 02, 2008, 12:15 pm
Duration: This information is not available in the database
Location: CAB G51
Speaker: Christian Lorenz Müller (Mosaic Group, ETH Zürich)
Many research questions in chemistry, biology and bioinformatics can be reformulated as search problems that are characterized by high dimensionality, non-convexity and noise. A prominent example is the protein folding problem (that is known to be NP-hard) that can be considered as an optimization problem over an energy landscape. In most case, such problems can only be tackled by stochastic search/optimization heuristics.
One of the most successful gradient-less optimization heuristic for non-convex, real-valued functions is the Evolution Strategy with Covariance Matrix Adaptation (CMA-ES). In the seminar we will present an extension of this algorithm, the Particle Swarm CMA-ES (PS-CMA-ES) algorithm that shows excellent performance on several hard benchmark problems. This heuristic can also be interpreted as a stochastic branch-and-bound algorithm.
In the outlook we will also give some insights about the interpretation of energy landscapes as weighted graphs, and the connection between protein folding and the MaxFlow-MinCut problem, and some open research topics. Hopefully, this seminar can stimulate future collaboration.
Automatic MiSe System Software Version 1.4803M | admin login