This Week’s Discoveries | 6 June 2017
- 6 juni 2017
Niels Bohrweg 2
2333 CA Leiden
- De Sitterzaal
Mast cells as effectors in cardiovascular disease
Ilze Bot (LACDR) is assistant professor in the BioTherapeutics cluster of the Leiden Academic Center for Drug Research, where she investigates the contribution of the mast cell to cardiovascular diseases.
Acute cardiovascular syndromes (ACS) are still the main cause of death in Western society. The main underlying pathology of ACS is atherosclerosis, which is caused by the accumulation of lipids and inflammatory cells in so-called atherosclerotic plaques. The mast cell, a potent immune effector cell type mainly recognized for its role in allergy and asthma, has been show to accumulate in the vessel wall during the progression of atherosclerosis. Previously, we have demonstrated that mast cells contribute to atherosclerotic plaque destabilization and recent data show that intraplaque mast cell numbers are of predictive value for future cardiovascular events in patients. In this project, we aim to determine how mast cells are activated in atherosclerosis and how we can prevent atherosclerotic plaque destabilization by targeting the mast cell.
Optimizing Evolutionary Optimizers
Thomas Bäck (LIACS) is professor of Natural computing at LIACS. In 2015, he received the prestigious IEEE Evolutionary Computation Pioneer Award for his contributions in synthesizing evolutionary algorithms. Thomas’ research interests are also in applications of data science and optimization to the life sciences, and in industrial applications in areas such as industry 4.0, process optimization, and product development.
In the field of evolutionary computation, the design and empirical validation of improvements of algorithms has always been a key area of interest. In this presentation, I will show how evolutionary algorithms can be used for this task.
After a short introduction to genetic algorithms and evolutionary strategies, the two main algorithm classes in this field, I will illustrate how modern evolutionary strategies can be significantly improved by automatically searching their algorithmic design space. Moreover, the results of this design space exploration can be analyzed using data mining to obtain additional insight into the contributions of certain algorithmic variations - an approach we are calling 'algorithm configuration data mining'.
An application to a medical image classification task will then illustrate the general applicability of evolutionary algorithms for configuring essentially any type of algorithms.
The talk is concluded by giving a brief explanation of the self-adaptation concept for mutation parameter learning in evolutionary algorithms.