Master's Project Defense: Louis Schlessinger

May 3
11:00 AM
1:00 PM
Jolley Hall, Room 309

Title: Automated Kernel Search Using Evolutionary Algorithms

Abstract: This project uses evolutionary algorithms to automatically select a kernel that explains a given fixed-size dataset well. Experiments compare different kernel search strategies and grammars on synthetic and real-world data. An algorithm is proposed that uses genetic programming to traverse the infinite search space of probabilistic models. This evolutionary kernel learning algorithm can discover underlying structure on a variety of objective functions.

Advisor: Roman Garnett