Abstract

Open distributed multi-agent systems are gaining interest in the academic community and in industry. In such open settings, agents are often coordinated using standardized agent conversation protocols. The representation of such protocols (for analysis, validation, monitoring, etc) is an important aspect of multi-agent applications. Recently, Petri nets have been shown to be an interesting approach to such representation, and radically different approaches using Petri nets have been proposed. However, their relative strengths and weaknesses have not been examined. Moreover, their scalability and suitability for different tasks have not been addressed. This paper addresses both these challenges. First, we analyze existing Petri net representations in terms of their scalability and appropriateness for overhearing, an important task in monitoring open multi-agent systems. Then, building on the insights gained, we introduce a novel representation using Colored Petri nets that explicitly represent legal joint conversation states and messages. This representation approach offers significant improvements in scalability and is particularly suitable for overhearing. Furthermore, we show that this new representation offers a comprehensive coverage of all conversation features of FIPA conversation standards. We also present a procedure for transforming AUML conversation protocol diagrams (a standard humanreadable representation), to our Colored Petri net representation.

Recent multi-agent systems (MAS) are built using an open, distributed design. These systems involve various challenges of monitoring geographically-distributed and independently-built multiple agents. Monitoring by overhearing [3] has been found to provide a powerful monitoring approach particularly suited for open distributed MAS settings. Here, an overhearing agent monitors the exchanged communications between the system's agents. It uses these observed communications to independently assemble and infer the needed monitoring information.Although overhearing can be used in many commercial and military implementations, the research in that direction has only been limited. Many previous investigations applied overhearing in context of specific applications leaving the general problem of overhearing unattained. In my research, I attempt to provide a comprehensive theoretical model for overhearing and then, based on this model, to systematically cover various aspects related to overhearing.

The Inductive Confidence Machine (ICM) provides an alternative method to that of the Transductive Confidence Machine (TCM) for complementing the bare predictions produced by traditional machine-learning algorithms with measures of confidence. These measures give an indication of how 'good' each prediction is, which is highly desirable in risk-sensitive applications. The motivation behind the introduction of the ICM was to produce algorithms that overcome the computational inefficiency problems suffered by TCMs.In this thesis, we study the ICM method, describing how it works and how it can be applied to different traditional machine-learning algorithms. More