Jan Tozicka

A framework for agent-based distributed machine learning and data mining

This paper proposes a framework for agent-based distributed machine learning and data mining based on (i) the exchange of meta-level descriptions of individual learning processes among agents and (ii) online reasoning about learning success and learning progress by learning agents. We present an abstract architecture that enables agents to exchange models of their local learning processes and introduces a number of different methods for integrating these processes. This allows us to apply existing agent interaction mechanisms to distributed machine learning tasks, thus leveraging the powerful coordination methods available in agent-based computing, and enables agents to engage in meta-reasoning about their own learning decisions. We apply this architecture to a real-world distributed clustering application to illustrate how the conceptual framework can be used in practical systems in which different learners may be using different datasets, hypotheses and learning algorithms. We report on experimental results obtained using this system, review related work on the subject, and discuss potential future extensions to the framework.

MALEF: Framework for distributed machine learning and data mining

Growing importance of distributed data mining techniques has recently attracted attention of researchers in multiagent domain. In this paper we present a novel framework MultiAgent Learning Framework (MALEF) designed for both the agent-based distributed machine learning as well as data mining. Proposed framework is based on: the exchange of meta-level descriptions of individual learning process; online reasoning about learning success and learning progress. This paper illustrates how MALEF framework can be used in practical system in which different learners use different datasets, hypotheses and learning algorithms. We describe our experimental results obtained using this system and review related work on the subject.

An abstract architecture for computational reflection in multi-agent systems

We propose a general framework for computational reflection in multi-agent systems and address some technical issues related to its implementation. Potentials of computational models of cognition and reflection in a multi-agent system are detailed, and using such models, an abstract architecture of a reflective agent is designed. We also propose important characteristics of reflective multi-agent systems build upon the presented architecture.

Reflective-Cognitive Architecture: From Abstract Concept to Self-Adapting Agent

This paper presents the general framework of the reflective-cognitive agent architecture. In our architecture we use modular approach to the reflection as we found it to be a promising way how to throw multi-agent systems together with the computing with limited resources, the programme code reusability and the automated code generation. The architecture is lightweight and it enables the agent to alter its own code in runtime using reflection according to the changes in the environment. At the end of the paper we present results of the architecture implementation showing the plausibility of the created prototype

Meta-reasoning for Agents’ Private Knowledge Detection

Agent’s meta-reasoning is a computational process that implements agent’s capability to reason on a higher level about another agent or a community of agents. There is a potential for meta-reasoning in multi-agent systems. Meta-reasoning can be used for reconstructing agents’ private knowledge, their mental states and for prediction of their future courses of action. Meta-agents should have the capability to reason about incomplete or imprecise information. Unlike the ordinary agents, the meta-agent may contemplate about the community of agents as a whole. This contribution presents application of the meta-reasoning process for the agent’s private knowledge detection within the multi-agent system for planning of humanitarian relief operations.

Utility-based model for classifying adversarial behaviour in multi-agent systems

Interactions and social relationships among agents are an important aspect of multi-agent systems. In this paper, we explore how such relationships and their relation to agentpsilas objectives influence agentpsilas decision-making. Building on the framework of stochastic games, we propose a classification scheme, based on a formally defined concept of interaction stance, for categorizing agentpsilas behaviour as self-interested, altruistic, competitive, cooperative, or adversarial with respect to other agents in the system. We show how the scheme can be employed in defining behavioural norms, capturing social aspects of agentpsilas behaviour and/or in representing social configurations of multi agent systems.