Miroslav Hudec
I am one of the PhD students at the Dept. of Cybernetics and Artificial Intelligence, Faculty of electrical engineering and informatics University of Technology in Kosice. I'm trying to make an research in area of the Artificial Intelligence in Artificial Neural Networks (google it!). I'm interested in online adaptive learning algorithms whose main aim is to act as a part of an entity in a dynamically changing environment.
Some of my research interests:
- Artificial Neural Networks - any kind of them
- Reinforcement Learning and it's combination with ANN's with the goal of creation of independent entity which will be able to act in anuncertain/unknown environment. To be able to act in such environment, one have to set up at least some kind of feedback, and if one want to be abstract from that env. it may to use only reinforcement (thus implementation RL into ANN's learning). Of course there are some theoretical issues whether the definition of reinforcement/punishment is an apriori information to acting entity (i think it is) and implementation challenges about - how to incorporate RL into distributed architecture (such as agents, neural networks), how to deal with (temporal) credit assignment problems and so on.
- Evolutionary Algorithms as general framework for case/(state) space exploration. However, because they are computationally expensive their applicability is limited to problems whose have their (cmputational) model.
Robotics - ufortunately many people look at robotics, as best the best testbed for above described approache(s). Yes, it has each property which an uncertain dynamic sytstem has to have, but always is there but... From control algorithm developer's point of view the repeatibility of experiment(s) is necessary for checking/validating the results. Moreover robotics is a field where different research fields meet, and therefore each part of the "control" loop have to be fully functional. My research is directed toward control/managing LegoMindstorms based robo-soccer team.
Personal Interests:
Life?
Contact:
Hudec Miroslav MSc.
Dept. of Cybernetics and Artificial Intelligence
FEI, Technical University of Kosice
Adress:
Letna 9
Košice, 04200
Slovak Republic
Phone: ++421-55-602-5101
E-mail: hudecm@neuron.tuke.sk
If u are exhausted by those boring and useless informations please take a look at my web folder. Warning: Informations contained in my web folder are only for educational purposes.
DISCLAIMER: Opinions presented on this page do not represent official opinions of CIT, CIG, TU of Kosice or any other organisation or group I am a memeber of.
