CS Forum: "Predictive modeling over evolutionary times" Indrė Žliobaitė, University of Helsinki

CS department's public guest lecture, open to everyone free-of-charge.

Indrė Žliobaitė
University of Helsinki

Host: Professor Jukka Suomela
Time: 14:15 (coffee at 14:00)
Venue: T3, CS building

Predictive modeling over evolutionary times

Abstract:

The problem of concept drift (changing data distribution) in machine learning is well acknowledged and many solutions exist for online adaptation. Those methods; however, are not straightforward to apply in reverse, where models built on the present day data are used for analyzing processes in the past. The talk will highlight computational challenges and transfer learning principles for dealing with changing data distribution in such scenarios. I will focus on computational methods for analysis of the global fossil and geological record, where time goes backwards millions of years, and the further back, the less is the correspondence between entities in the present and in the past.

Bio:

Indrė Žliobaitė is an Assistant Professor at the Department of Computer Science, University of Helsinki, where she leads a research group on data science and evolution, focusing on computational methods for understanding macroevolutionary processes and biospheric change.