We focus on research projects in the fields of:
Clustering is a fundamental task in machine learning and data analysis. One drawback of spectral clustering - a state-of-the-art method consistently outperforming classical methods - is its computational complexity and thus limited applicability to large-scale problems. This article presents a new scaling approach that identifies a small set of representative data points on which to perform the clustering. The approach obtains high-quality solutions for data sets with up to 20,000 data points within a few seconds.
Read our IEEM-paper to learn more.
Managers often use project-management software to schedule projects and allocate scarce resources to different activities. In several software packages, the user can influence the resource-allocation procedure by selecting a rule for computing the priority values of the various activities. We experimentally evaluate the resource-allocation methods of eight widely-used software packages and identify effective priority rule selection strategies.
Read our FSMJ-paper to learn more.
