http://popups.lib.uliege.be/1373-5411/index.php?id=1136 Index terms fr 0 http://popups.lib.uliege.be/1373-5411/index.php?id=3014 Artificial agents societies are well suited to design and implement open distributed systems. As the complexity of such systems grows, the design of agent societies with a complete pre-defined behaviour is a significant challenge due to the dynamic interactions among agents and between agents and the environment. To overcome existing difficulties, agent systems with emergent behaviour are a fertile area of research and span a large range of applications. The paper presents an analysis of the anticipatory capabilities of multi-agent systems endowed with emergent behaviour, by considering both reactive multi-agent systems and cognitive multi-agent systems. Our approach in analysis is driven by the different levels on which predicted behaviour can be achieved and is illustrated by two scenarios, one for each case of agent system, grounding the view for the cognitive case in our previous related work on self-organizing systems with cognitive agents. Fri, 06 Sep 2024 15:59:38 +0200 Thu, 10 Oct 2024 10:10:18 +0200 http://popups.lib.uliege.be/1373-5411/index.php?id=3014 http://popups.lib.uliege.be/1373-5411/index.php?id=2689 When designing information systems, it would be good to be able to compare alternatives. However, information systems are complex phenomena as they encompass the humans involved in distributing the information. One possible way of making comparisons would be through simulation. Having constructed a prototype for such a simulation we have seen that the traditional approaches, such as Cellular Automata utilized within the social simulations field are usable but not sufficient. However, the newer agent-based approaches show more promise. We conclude that in order to make simulations of our kind possible, the new technologies, such as multi-agant systems, need be adapted and extended. One of the pieces missing is an agent-based infrastructure building on anticipatory principles for agent information behavior. Fri, 30 Aug 2024 11:09:22 +0200 Mon, 07 Oct 2024 15:49:09 +0200 http://popups.lib.uliege.be/1373-5411/index.php?id=2689 http://popups.lib.uliege.be/1373-5411/index.php?id=2839 The paper presents a model of heuristic negotiation between self-interested agents which allows the use of arguments, negotiation over multiple issues of the negotiation object, single and multi-party negotiation, and learning of the agent's negotiation primitives. The model uses negotiation objects and negotiation frames to separate the object of negotiation from the negotiation process. In order to negotiate strategically, the agents use a reinforcement learning algorithm applied on a specific state space representation of the negotiation process. Tue, 03 Sep 2024 15:20:24 +0200 Tue, 03 Sep 2024 15:21:51 +0200 http://popups.lib.uliege.be/1373-5411/index.php?id=2839 http://popups.lib.uliege.be/1373-5411/index.php?id=1132 In order to build adaptive artificial systems, we suggest a thesis about the adequacy of a system relative to its relationships with the environment. We can tell it in this way: "Any system having a cooperative internal medium is functionally adequate". In our learning method, a global function emerges from the system unless this function is not explicity dictated in each of its component. In fact, each component is always looking for maintaining a cooperative situation with others components in the system. The originality of our method is its facility of implementation because it's totally independent of the application field. Another property is that our method allows to suppress the classical component of control : the cooperation permits it implicitly. Fri, 05 Jul 2024 14:19:11 +0200 Fri, 05 Jul 2024 14:19:21 +0200 http://popups.lib.uliege.be/1373-5411/index.php?id=1132