The website of Albert van der Heide.

this website

Welcome to the website of Albert van der Heide. The information that you can find here is my curriculum vitae, current work, and contact information. For questions or any other matter, please feel free to contact me.

Current position

Currently I am working for DuPont. Here I fulfill the position of Global Mobility Coordinator and I support the Staffing team.

Ph.D., Computer Science, Artificial Intelligence

Degree of Doctor by University of Oviedo - Programa Oficial de Doctorado en Ingeniería Informática. Cum laude.
Topic: Affect and Emotion in Artificial Agents.

Automatic recognition of human emotions, as well as the expressivity of affect and emotions in artificial agents, has advanced considerably over the last decade or so. The recognition and expression of emotions is used in artificial agents to make these appear more lifelike, and appears on first sight to be quite advanced already. However, the main unresolved question for artificial agents remains how to create internal affect and emotion dynamics that make it possible for the agent to react and respond with appropriate emotions at the right time, in a convincing and believable way. This topic has been central in my dissertation.
One of the first issues to confront when dealing with concepts such as "feelings", "emotions", and "moods", is that one discovers that these are poorly defined. From everyday use in language and folk psychology one would like to believe that their meaning is clear, however when formulating definitions researchers differ considerably in used conceptualizations. This is true in psychology and even more so in computer science.

Affect and Emotion in Psychology. A study has been performed of how affect and emotions are conceptualized in psychology, which has led to the use of a clear working definition of the otherwise fuzzy concepts of emotion, mood, and affect. As a result of the study performed, we have adopted the conceptualization of emotions as proposed by James Russel. At the core of all affective feelings and emotions are bodily sensations. These bodily sensations are conceptualized as Core Affect, which consist of two dimensions: the first dimension being sensations of unpleasantness / pain versus pleasantness (valence), the second dimension being feelings of passiveness versus activation (arousal). Core Affect feelings can exist in consciousness without having been labelled, without having been given a name. It is when these feelings are given a name by consciousness that emotions or moods are constructed, for example by labelling the feelings as "I feel angry (because of what someone did)", "I feel love (towards someone)", "I feel shame (because of what I did)", etc. Thus, emotions are an explanation generated by consciousness, a psychological construction, and in this generated explanation the mind takes into account not only the perceived external world, but also the currently experienced internal feeling of Core Affect.

Affect and Emotion in Computer Science. At the same time a review of currently available computational models for simulation of affect and emotions has been made. This study has been twofold: we have evaluated if the currently available models use a clear conceptualization of what affect and emotions are, and at the same time identify the existing difficulties in current computational models of affect and emotion, for example, deterministic and predicable agent behavior and the difficulty in dealing with conflicting emotions.

Contributions. Based upon these findings we have proposed a general architecture for affect and emotion capable agents. This architecture demonstrates how a non-affect capable agent can be extended with "affect dynamics" (Core Affect) and "reasoning about affect" (cognitive construction of emotions). Next, we developed the Core Affect model, including its internal dynamics and how it can interact with other more standard agent modules for processes such as perception, thinking and action. Using this conceptualization, Core Affect functions as an affective memory and its dynamics is at the core of the agent's affective life.
To assist designers into bringing the proposed architecture and core affect model to real world applications, we have developed a methodology for affective agent creation. The methodology is elaborate and complete. Among others, it provides a means to go from high level personality trait definitions to concrete model parameter settings for affective life simulation, and several methods to realize emotion construction are discussed. The methodology is demonstrated and discussed with practical hands-on examples from an illustrative use-case scenario.

M.Sc., Computer Science, Artificial Intelligence

Degree of Master in Science by University of Groningen - Faculty of Psychology, Education and Sociology.
Topic: The group movement of the Aptenodytes Forsteri (emperor penguin) implemented as an "emergent" phenomenon.

Emperor Penguins live in the artic in an extremely harsh environment. To survive the freezing blizzards the penguins group together to minimize exposure to the cold. The penguins rotate and take turns standing on the outside of the group and no individual penguin can survive continuous exposure to the cold on the outside. Actually, for the survival of the group as a whole this turn-taking behaviour is absolutely necessary. However, it is highly unlikely that this beneficial group behaviour is caused by conscious and deliberate turn-taking of the individual penguins. To the contrary, it much more likely that this beneficial group behaviour is simply caused by the interaction of many egoistically motivated individual penguins. In other words, the beneficial group behaviour is an "emergent" property of the system.
The objective of the M.Sc. thesis has been to reproduce the beneficial rotating group behaviour of the penguins by modelling and programming simple "egoistically motivated" penguin agents. As such we consider this to be a self-organizing system. The M.Sc. was developed within the context of "swarm intelligence", a field which studies how groups of basic and simple agents together can demonstrate new and more complex macroscopic behaviours or properties.