Towards Natural Language Understanding using Multimodal Deep Learning

This thesis describes how multimodal sensor data from a 3D sensor and microphone array can be processed with deep neural networks such that its fusion, the trained neural network, is a) more robust to noise, b) outperforms unimodal recognition and c) enhances unimodal recognition in absence of multimodal data. We built a framework for a complete...

Deep Reinforcement Learning - Pretraining actor-critic networks using state representation learning

In control, the objective is to find a mapping from states to actions that steer a system to a desired reference. A controller can be designed by an engineer, typically using some model of the system or it can be learned by an algorithm. Reinforcement Learning (RL) is one such algorithm. In RL, the controller is an agent that interacts with the...

Analyzing and applying agent oriented programming methods for teaching purposes

In this thesis we discuss ways to improve teaching methods for agent oriented programming. The research approach consists of several steps. First data is gathered from groups of students participating in a first year agent programming project that uses the Unreal Tournament 2004 environment. The stronger and weaker groups are analyzed in a...

Learning the Model Structure of Dynamic Bayesian Networks for Automated Speech Recognition

Improving the performance of Automated Speech Recognition system requires incorporating more knowledge in the model of Automated Speech Recognition system. Information such as the context of the conversation and the characteristics of the speaker can make the task of recognizing speech more accurate. The challenge is how this knowledge can be...

The developing of the foundation for the Monitoring and Diagnosing System

A start has been made implementing the MDS based on an Expert System with the recommended Progressive Reasoning strategy and laying the foundation for further developments. An interrupt handling is introduced that reduces the time the MDS is interrupted in its diagnosing task and therefore able to find more refined conclusions.

Een Semantisch Connectionistisch Redeneersysteem (SCORE)

Het onderzoek heeft zich gericht op de ontwikkeling van een Semantisch Connectionistisch Redeneersysteem (SCORE). SCORE kan de basis vormen voor een volwaardig expertsysteem. De, voor een expertsysteem van elementair belang zijnde, fundamentele inferentie principes, kennisacquisitie principes en kennisrepresentatievormen hebben in SCORE de...