Speaker:

Lech Szymanski - Department of Computer Science

Title:

Deep architectures and classification by intermediary transformations

Location:

Archway 2 - 1:00 pm, Friday 6 May

Abstract:

In this talk, I will examine the learning methods for machine classification with a focus on support vector machines and multi-layer feed forward neural network models. I will present a theory that defines a supervised objective for layered training of deep architectures and a classification algorithm based on the new theory. Results from a range of different classification problems will be discussed.