< Back to previous page

Project

Induction of multi-directional ensembles of decision trees

The field of data mining covers a large variety of tasks, such as classification, regression, clustering, outlier detection, subgroup discovery, probabilistic inference, …, and an even larger variety of methods.  Many methods have a high computational complexity, which makes them unsuitable for analyzing very large datasets.  A possible solution to this is the automatic construction, from data, of a model that can next be analyzed instead of the data itself.  Ideally, the construction of such a model takes time linear in the data size, and analysis of the model takes time independent of the data size (“instantaneous mining”).  In this project, we aim to investigate to what extent a novel model class, multi-directional ensembles of decision trees, fits this description.  The research will focus on the efficient construction of these models from data, in such a way that versatility of the model is guaranteed.

Date:4 Oct 2016 →  4 Oct 2020
Keywords:Data Science, Machine Learning, Big Data
Disciplines:Applied mathematics in specific fields, Computer architecture and networks, Distributed computing, Information sciences, Information systems, Programming languages, Scientific computing, Theoretical computer science, Visual computing, Other information and computing sciences
Project type:PhD project