This is a past event
Direct evolution of process algebra model parameters
Abstract: Process algebras are an effective method for defining models of complex interacting systems, especially biological systems, but tuning parameters to allow model outputs to match experimental data can be difficult. This is the well-known parameter fitting problem. Evolutionary algorithms are powerful methods for finding solutions to optimisation problems with large search spaces, such as the parameter fitting problem mentioned. In this talk I'll present a framework bringing together evolutionary computation techniques with modelling using process algebra to provide numeric parameters for predefined models. The tuned models can then be used confidently for further simulation or analysis. Moreover, further insight into the system under investigation may be gained by examining the performance of the evolutionary algorithm. The Evolving Process Algebra (EPA) framework will be demonstrated through benchmark examples from systems biology or computer science. If there's time, I will also talk about the next step in our work, which is evolving the models themselves, not just the parameters. This is joint work with David Cairns and David Marco (also at Stirling).
- Speaker
- Carron Shankland
- Hosted by
- George Coghill
- Venue
- MT203