Robin Lamarche-Perrin

How to Build the Best Macroscopic Description of Your Multi-Agent System?

The design and debugging of large-scale MAS require abstraction tools in order to work at a macroscopic level of description. Agent aggregation provides such abstractions by reducing the complexity of the microscopic description. Since it leads to an information loss, such a key process may be extremely harmful for the analysis if poorly executed. This paper presents measures inherited from information theory to evaluate abstractions and provide the experts with feedback regarding the quality of generated descriptions. Several evaluation techniques are applied to the spatial aggregation of an agent-based model of international relations. The information from on-line newspapers constitutes a complex microscopic description of agent states. Our approach is able to evaluate geographical abstractions used by the domain experts in order to provide efficient and meaningful macroscopic descriptions of the world global state.

The Best-Partitions Problem: How to Build Meaningful Aggregations

The design and the debugging of large distributed AI systems require abstraction tools to build tractable macroscopic descriptions. Data aggregation provides such tools by partitioning the system dimensions into aggregated pieces of information. Since this process leads to information losses, the partitions should be chosen with the greatest caution. While the number of possible partitions grows exponentially with the size of the system, this paper proposes an algorithm that exploits exogenous constraints regarding the system semantics in order to find the best partitions in a linear or polynomial time. Two constrained sets of partitions (hierarchical and ordered) are detailed and applied to spatial and temporal aggregation of an agent-based model of international relations. The algorithm succeeds in providing meaningful high-level abstractions for the system analysis.

Evaluating trace aggregation for performance visualization of large distributed systems

Performance analysis through visualization techniques usually suffers semantic limitations due to the size of parallel applications. Most performance visualization tools rely on data aggregation to work at scale, without any attempt to evaluate the loss of information caused by such aggregations. This paper proposes a technique to evaluate the quality of aggregated representations - using measures from information theory - and to optimize such measures in order to build consistent multiresolution representations of large execution traces.

A spatiotemporal data aggregation technique for performance analysis of large-scale execution traces

Analysts commonly use execution traces collected at runtime to understand the behavior of an application running on distributed and parallel systems. These traces are inspected post mortem using various visualization techniques that, however, do not scale properly for a large number of events. This issue, mainly due to human perception limitations, is also the result of bounded screen resolutions preventing the proper drawing of many graphical objects. This paper proposes a new visualization technique overcoming such limitations by providing a concise overview of the trace behavior as the result of a spatiotemporal data aggregation process. The experimental results show that this approach can help the quick and accurate detection of anomalies in traces containing up to two hundred million events.

Analysis of International Relations through Spatial and Temporal Aggregation

A leading topic of the GEOMEDIA project [1] focuses on the analysis of international relations through print media. In this demo paper, we show how techniques for Multi-Agent Systems aggregation [2,3] can be applied to the spatial and temporal aggregation of news. Two experiments show that the generated multi-resolution representations may draw attention to critical areas for the analysis.

Agrégation de traces d’exécution pour la visualisation de grands systèmes distribués

RÉSUMÉ. La visualisation de performance consiste à représenter graphiquement l’exécution d’applications parallèles pour procéder à leur analyse. Dans le cas de très grands systèmes, l’agrégation des données analysées est inévitable. Cet article met en évidence un problème scientifique majeur : comment produire des représentations agrégées qui ont un sens lors du passage à l’échelle ? Quatre éléments de réponse sont discutés. (1) Les outils de visualisation doivent garantir la compréhension et le contrôle par l’utilisateur du procédé d’agrégation. (2) Il est crucial d’estimer la qualité des représentations engendrées afin de distinguer les agrégations utiles (suppression d’informations redondantes) de celles qui sont dangereuses pour l’analyse (perte d’informations importantes). Nous proposons des mesures issues de la théorie de l’information pour quantifier ces deux aspects et pour ainsi engendrer des représentations multirésolutions (agrégation de données redondantes et conservation des données hétérogènes). (3) Les propriétés sémantiques et topologiques du système doivent également être prises en compte afin de garantir la juste interprétation des agrégats par l’utilisateur. La recherche des agrégations optimales est ainsi restreinte à un ensemble d’agrégations pertinentes sur le plan sémantique. (4) Des méthodes de calcul sophistiquées sont alors nécessaires pour sélectionner les agrégations optimales lors du passage à l’échelle. Nous proposons un algorithme à complexité linéaire (dans le cas de systèmes hiérarchiques) parvenant à agréger jusqu’à un million d’entités et garantissant l’interprétabilité des visualisations engendrées.