WAI › Schedule › 2018 › 26 nov

Title : ‘Design of Adaptive Temporal-Causal Network Models for Handling Extreme Emotions’

Presenter Sahand Mohammadi Ziabari
Abstract In recent literature from Neuroscience the adaptive role of the effects of stress on decision making is highlighted. The problem addressed in lecture is how that can be modelled computationally. The adaptive effect of acute severe stress on decision making is addressed based on a Network-Oriented Modeling approach. The presented adaptive temporal-causal network models addresses the therapies, like music, humor, and drugs, which have been used to decrease the stress level of individuals with post-traumatic stress disorder (PTSD) and also the suppression of the existing network connections in a first phase as a result of the acute stress, and then as a second phase relaxing the suppression after some time and give room to start new learning of the decision making in the context of the stress again.

Title : Using Genetic Programming to Generate Dynamical Systems Models for Health Care

Presenter Mark Hoogendoorn
Abstract The huge wealth of data in the health domain can be exploited to create models that predict development of health states over time. Temporal learning algorithms are well suited to learn relationships between health states and make predictions about their future developments, however these models: (1) either focus on one generic model for all patients, providing general insights but often with limited predictive performance, or (2) are individualized models from which it is hard to derive generic concepts. In this presentation, I will present a middle ground, namely parameterized dynamical systems models that are generated from data using a genetic programming (GP) framework we introduce. A fitness function suitable for the health domain is defined. An evaluation of the approach in the mental health domain shows that performance of the model generated by the GP is on par with a dynamical systems model developed based on domain knowledge, significantly outperforms a generic LSTM model and in some cases also outperforms an individualized LSTM model.