Academy projects will tackle heterogeneous big data, genome research, and tensor-based machine learning

 

The department of Computer Science was successful in the recent call of the Academy of Finland. Three four-year projects funded by the Academy will run from 9/2017 until 8/2021. The principal investigators of the projects are Professor Veli Mäkinen and Associate Professors Jiaheng Lu and Teemu Roos, respectively.

 

Prof Lu's project Holistic Query Optimization and Transaction Processing in Multi-model Data Management will address the challenge of  the variety in data. The project proposes to develop an approach to handle heterogeneous large scale data management via introducing a unifying data storage, access and management model, indexing techniques for query optimization and a novel approach to managing consistency.

 

Prof Roos's group is partnering with Assistant Professor Tapio Pahikkala at the Department of Information Technology, University of Turku. The consortium project's title is Tensor-Based Machine Learning for Big Data with Inherent Dependencies. Its goal is to develop computationally and statistically advanced machine learning methods in situations that are not ideally captured by the standard setting where data are assumed to be independently sampled. Applications include so called "zero-shot learning" in drug-target interaction prediction.

 

Prof Mäkinen's project Sequence analysis revisited and extended studies implications of bidirectional Burrows-Wheeler transform -based techniques to sequence analysis and widens the scope of such analyses to labeled directed acyclic graphs (labeled DAGs). The motivation for these new directions come from several timely applications in genome research, e.g., in new ways to represent diploid genomes and pan-genomes in place of the commonly adopted use of a single sequence as the basis of analyses.

 

More information: