Towards Emotion Recognition using Evolutionary Computation

Facial expression recognition and analysis are difficult because huge amounts of input data needs to be processed and automation is even more complex. However, Evolutionary Computing (EC) can be good at complex tasks. For implementation, Evolvable Hardware (EHW) is an advantageous technology which is fast in real-time situations. This thesis uses Evolutionary Computing (EC) for facial expression analysis, and proposes new algorithms to undertake a limited set of facial expression analysis. Compared with conventional classifiers such as Support Vector Machines ( SVMs ), results of using a Cartesian Genetic Programming ( CGP ) classifier show better effectiveness when using the technique. Finally, this thesis discusses Evolvable Hardware (EHW) but does not implement it.