AI-MAS Laboratory        Computer Science and Engineering Department        Faculty of Automatic Control and Computers        University Politehnica of Bucharest


Mihaela-Alexandra Puica

Eng. Mihaela-Alexandra Puica, PhD Student

Superviser: Prof. Adina Magda Florea
PhD fundamental domain: Engineering Science
PhD domain: Computer Science and Information Technology
Duration: 2010-2013
AI-MAS Laboratory
Computer Science and Engineering Department
Faculty of Automatic Control and Computers
University "Politehnica" of Bucharest
contact: mihaela(dot)puica(at)gmail(dot)com



Research

My current research is done in the context of my doctoral studies, whose subject lies in the field called affective computing. That is, I am interested in understanding the mechanisms of emotions, how they appear and how they influence human behavior, so that I can apply these techniques to build intelligent agents.

There is this common belief that "emotional" is synonym with "irrational" and rationality has nothing to do with emotions. While I agree that too much emotion can impede clear judgement, several studies have proven that too little emotion also impairs decision making. Therefore, an optimum level should be reached, although it isn't yet very clear how to measure or evaluate the quantity of emotion a person feels. However, we must admit emotion and cognition are interdependent, so why should we separate them when developing artificial intelligent agents?

Emotions influence all cognitive processes like perception, learning and memory. They have a major role in decision making by narrowing the possibilities considered and, by associating events with positive or negative emotions, one can learn what could be the best decision in a given situation - but usually they are not aware of this mechanism, so they call it a "gut feeling". Moreover, some situations require more resources to be dealt with, and a certain amount of emotional arousal is able to release them (even though one isn't even aware that they existed). All in all, emotions are a key to improved performance.

For agents to be "humanly intelligent", they should be able both to recognize user emotions and to simulate their own emotions (display emotional behavior). These are two different research paths in the affective computing field. My concern is with emotion simulation: I am developing a computational model of emotions that will enable my agent (built based on this model) to behave intelligently. The focus is on resource usage and improved performance, but human-computer interaction shouldn't be ignored: an emotional agent that understands the users' affective state and is able to empathize with them is a more pleasant companion then a rational, but non-emotional agent.



PhD

Thesis (en): Towards a computational model of emotions for enhanced agent performance

Summary (ro): Catre un model computational al emotiilor pentru imbunatatirea performantelor agentilor


Research reports
(Program de cercetare stiintifica)

1. Affective techniques for emotion simulation
2. Towards a model of agent affective behavior
3. PhD Thesis Proposal


Exams
(Program de pregatire universitara avansata)

1. Internet Information Retrieval Methods (Metode de regasire a informatiei in Internet) - January 24th, 2011
2. Developing Complex Applications in Java (Dezvoltarea aplicatiilor complexe in Java) - February 3rd, 2011
3. E-learning Systems (Sisteme de e-learning) - September 15th, 2011
4. Affective Computing and Emotional Agents (Calcul afectiv si agenti emotionali) - September 29th, 2011



Education

2010-2013 - PhD Student, Computer Science and Engineering Department, Faculty of Automatic Control and Computers, University Politehnica of Bucharest

2009-2011 - M.Sc. in Artificial Intelligence, Computer Science and Engineering Department, Faculty of Automatic Control and Computers, University Politehnica of Bucharest

2004-2009 - B.Sc. in Computer Science, Computer Science and Engineering Department, Faculty of Automatic Control and Computers, University Politehnica of Bucharest

2000-2004 - Mathematics-Informatics, "Mihai Viteazul" National College, Bucharest


This work is supported by University "Politehnica" of Bucharest and POSDRU Grant No. POSDRU/107/1.5/S/76813.