Hi! Let's meet up again. This time we have two presenters: Roger will tell us a bit about spatial data analysis in R, and Tomasz will present his work on genetics and cancer diagnostics using deep learning methods. It is going to be really interesting to hear what these two experienced academics have to tell us.
Please be punctual.
Our sponsors for this meetup are DNB and Sonat Consulting.
Hope to see you there!
16:45 - We meet for some food and mingling
17:15 - Introduction and information about the meet-up
17:30 - "Applied spatial data analysis with R: status and prospects"
Roger Bivand, PhD, is a professor at the Norwegian School of Economics. His current research interests are in developing and supporting the development of open source software for analysing spatial data, including spatial econometrics. He has been active in the R community since 1997, and is now an auditor of the R Foundation. He is an editor of the Journal of Statistical Software, the R Journal, Journal of Geographical Systems, Geographical Analysis and Norsk Geografisk Tidsskrift. He is often invited to teach external PhD courses in applied spatial data analysis.
Roger will introduce us to some of the concepts of, and available tools to handle, spatial and spatio-temporal data in R:
"Spatial and spatio-temporal data are characterised by structures that distinguish them from typical tabular data. The geometric structures also have spatial reference system information, and can adhere to standards, which may ease geometrical operations.
Spatial data analysis can very fast become computationally intense. Such tasks include interpolation, upsampling, focal operations, change of support and handling vector data with very detailed boundaries, as well as modelling using Bayesian inference.
A further challenge to modelling using training sets with spatial data is how to split the observations in the presence of spatial dependence."
18:10 - Break
18:20 - "Applications of deep learning in medical genetics and cancer diagnostics"
Tomasz Stokowy, PhD, has a MSc in automatic control and robotics and a PhD, both from the Silesian University of Technology, Gliwice, Poland. He completed postdoctoral fellowship "Deep sequencing in biomedicine" at the University of Bergen. During his postdoctoral fellowship was affiliated with Yale School of Medicine.
Tomasz currently works in Genomics Core Facility at the University of Bergen. His research is focused on whole human genome analytics and interpretation. He is also involved in development of bioinformatics tools and molecular diagnostics of genetics related disorders.
Tomasz will be presenting his work on genetics and cancer diagnostics using deep learning methods.
"Human genome consists of[masked] nucleotides. We know that some of them are essential for our lives, but the function of many is still unknown. Development of new IT methods allows us to understand human genome better, provide patients with accurate diagnosis and choose the most suitable, personalized treatment options.
Our benchmarks confirm, that deep learning algorithms outperform classic, CPU based methods of genome analysis. Therefore I introduced TensorFlow based algorithms to Haukeland Hospital, accelerated them by GPUs and standardized using Docker technology.
Finally, I will show how deep learning can be used to fight rare disorders and cancer, including examples of cases from the clinic."
19:00 - Thank you and see you next time :)