ETRO VUB
About ETRO  |  News  |  Events  |  Vacancies  |  Contact  
Home Research Education Industry Publications About ETRO

Master theses

Current and past ideas and concepts for Master Theses.

Nonlinear Prediction of Emotion by Reinforcement Learning

Subject

Important developments of new sensing methods and pattern recognition techniques have motivated an increasing research in the field of affective computing, a recent but fast evolving research discipline which focuses on producing applications that adapt to the user’s emotions. In this context, affect detection turns to be the key to building systems that automatically respond to a user’s affective state in order to enhance the quality of the interaction.
Emotions perform a significant part in personal or social communication and can be conveyed verbally by expressive words or expressed by non-verbal signs such as facial expressions and gestures. Most of the modern Human Computer Interaction (HCI) platforms are lacking in translating the human emotional states to derive the right actions to execute. The main goal of this project is to fill this gap by efficiently detecting emotional states that can help to enrich the HCI systems in the 21st century.

Kind of work

The most recently used approach for emotion modelling is the dimensional approach describes the affective state of a person by certain continuous attributes (dimensions). The most widely used dimensional model is the Circumflex of Affect [1] and its dimensions are valence and arousal. Arousal describes how intense is the emotional experience, while valence refers to the level of pleasure related to an emotion, and takes positive and negative values for pleasant and unpleasant emotions, respectively. Several approaches have been proposed for continuous prediction of emotions in time series based Valence –Arousal, among them Recurrent Neural Networks (RNN) have been proposed as efficient methods for prediction [2]. Meanwhile, reinforcement learning, a kind of goal-directed learning, is of great use for a learner (agent) adapting unknown environments [3]. This work aims at constructing a RNN Neural Networks combined with Reinforcement Learning [4] and their Applications to emotion prediction.

Framework of the Thesis

[1] J. A. Russell, “A circumplex model of affect,” Journal of personality and social psychology, vol. 39, no. 6, 1980, pp. 1161–1178.
[2] M. Wollmer, F. Eyben, S. Reiter, B. Schuller, et al. “Abandoning emotion classes-towards continuous emotion recognition with modelling of long-range dependencies”. In INTERSPEECH, 2008, pp. 597–600.
[3] Sutton, R.S., Barto, A.G., Reinforcement Learning: An Introduction. The MIT Press (1998).
[4] Takashi Kuremoto, Masanao Obayashi, Kunikazu Kobayashi, “Nonlinear Prediction by Reinforcement Learning”, Advances in Intelligent Computing, Volume 3644 of the series Lecture Notes in Computer Science, 2005, pp 1085-1094

Expected Student Profile

• Following an MSc in a field related to one or more of the following: electrical engineering, computer science, applied mathematics, Biomedical Engineering
• Experience with machine learning and statistics.
• Experience with image processing and signal processing.
• Strong programming skills.
• Ability to write scientific reports and communicate research results at conferences in English.

Promotor

Prof. Hichem Sahli

+32 (0)2 629 2916

hsahli@etrovub.be

more info

Supervisor

Dr. Meshia Oveneke

+32 (0)2 629 2969

mcovenek@etrovub.be

more info

- Contact person

- IRIS

- AVSP

- LAMI

- Contact person

- Thesis proposals

- ETRO Courses

- Contact person

- Spin-offs

- Know How

- Journals

- Conferences

- Books

- Vacancies

- News

- Events

- Press

Contact

ETRO Department

info@etro.vub.ac.be

Tel: +32 2 629 29 30

©2019 • Vrije Universiteit Brussel • ETRO Dept. • Pleinlaan 2 • 1050 Brussels • Tel: +32 2 629 2930 (secretariat) • Fax: +32 2 629 2883 • WebmasterDisclaimer