On the 10th of May, I will be at the Argumentation in Multi-Agent Systems (ArgMAS) workshop, which is held as part of this year’s Agents and Multi-Agent Systems (AAMAS) conference. I will present the work from my accepted paper ‘A Formal Argumentation Framework for Deliberation Dialogues’, which was written together with Henry Prakken, Gerard Vreeswijk and John-Jules Meyer.
Agents engage in deliberation dialogues to collectively decide on a course of action. To solve conflicts of opinion that arise, they can question claims and supply arguments. Existing models fail to capture the interplay between the provided arguments as well as successively selecting a winner from the proposals. This paper introduces a general framework for agent deliberation dialogues that uses an explicit reply structure to produce coherent dialogues, guides in outcome selection and provide pointers for agent strategies.
This month I started my new job as a PhD student at the Intelligent Systems group at the Utrecht University. I will be working on the NWO-funded project ‘Agents Interacting in Dialogues with Argumentation’, or AIDA. We will try to prove if and how the use of argumentation in agent to agent dialogue can be beneficial and how this is measured. My main supervisor is Prof. Mr. Dr. Henry Prakken.
In two weeks I will give a presentation on the Bos22apl project. It is titled ‘Adaptive reinforcement learning agents in real-time strategy games’ and it will explain the project goals, the design of the research platform and how it is implemented and discusses some preliminary results. The presentation can also be downloaded for anyone interested.
I also updated the Bos22apl page, which will be the place that contains the most updated information, lists all downloads, etc. Direct link: http://ekok.nl/tech/bos22apl
I’m currently busy with my Master’s degree graduation project on machine learning in computer games. This video shows some preliminary work on learning a counter-strategy in the Bos Wars RTS game using 2apl.
For my study I did an experimentation project on learning in agent environments. The idea is to build action selection learning into the 2APL agent platform. It uses different reinforcement learning techniques to allow software agents to learn what actions give the best result while trying to reach a goal.
The full report can be downloaded and describes how it was implemented and what RL methods worked best in a random gridworld environment searching for bombs.
Last Monday I gave a presentation on the philosophical subject ‘Can computers be creative?’. Everything went well and the discussion was lively. The PowerPoint presentation can be download here. It covers topics like the definition of creativity and if artificial creativity can be build in principle and it shows some practical research projects done on this subject.
A very interesting and accessible discussion transcript between Kurtzweil, Gelernter and Brooks can be found at KurtzweilAI.net: “Will machine become consciousness?”.