Supervisors:
Working thesis:
Optimal Decision Trees for Ensemble Learning.
Typically, Decision Trees are made using quick, greedy algorithms that don’t guarantee that the resulting tree is optimal. Optimal Decision Trees address this issue by using a more complex Mixed Integer Optimisation algorithm that attempts to find the best tree of a given structure.
Ensemble learning involves combining many models to make a prediction. Each model makes its own prediction, and the results are aggregated so that every prediction is considered, which produces an overall model that is more accurate than any individual model. This thesis investigates the effectiveness of using ensembles composed of Optimal Decision Trees for classification tasks by analysing how using different versions of the tree impacts the accuracy of the ensemble, and comparing these ensembles to well-known ensemble methods.
Research interests:
Machine learning, Deep learning
Academic history:
Bachelor of Science in Statistics
