MSc Thesis Presentation: Semi-supervised learning

[Jesper]

Hi all,

I'm on the verge (or so I hope) of completing my Master's thesis, and that means I've got my presentation coming up. I'd hereby like to invite all of you to join this fun event. My thesis was about machine learning, more specifically semi-supervised ensemble learning, and the talk will be pretty accessible to anyone with a computer science background and an interest in machine learning.

Time: Thursday April 26th, 11.00
Location: Room 402, Snellius
Duration: +- 45 minutes (including questions)

Below this message, there's an outline of the talk and more information.

Cheers!
Jesper van Engelen

----------

Title: Semi-supervised ensemble learning
Student: Jesper van Engelen
Supervision: Holger Hoos & Matthijs van Leeuwen

Outline:
In many machine learning problems, labeled data is scarse or expensive to obtain, whereas unlabeled data is abundant and cheap. Semi-supervised learning is the branch of machine learning that considers using labeled and unlabeled data jointly to build better learners. In this thesis presentation and defense, we propose a new semi-supervised ensemble learning method based on automated machine learning. We achieve state-of-the-art performance in multiclass classification problems using AutoML. Furthermore, we examine the field of semi-supervised learning and, in particular, semi-supervised classification.

The talk will provide a gentle introduction to semi-supervised learning and is suited for anyone with a background in computer science (or a related discipline) who is interested in machine learning.