文档介绍：Data Clustering and Visualization using
Cellular Automata Ants
Andrew Vande Moere, Justin J. Clayden, and Andy Dong
Key Centre of puting and Cognition
The University of Sydney, Australia
{andrew, justin, adong}***@
Abstract. This paper presents two novel features of an emergent data
visualization method coined “cellular ants”: unsupervised data class labeling
and shape negotiation. This method merges characteristics of ant-based data
clustering and cellular automata to plex datasets in meaningful
visual clusters. Cellular ants demonstrates how a decentralized multi-agent
system can autonomously detect data similarity patterns in multi-dimensional
datasets and then determine the according visual cues, such as position, color
and shape size, of the visual objects accordingly. Data objects are represented
as individual ants placed within a fixed grid, which decide their visual attributes
through a continuous iterative process of pair-wise localized negotiations with
neighboring ants. The characteristics of this method are demonstrated by
evaluating its performance for various benchmarking datasets.
1 Introduction
This paper proposes a simple approach towards unsupervised data visualization. It
uses principles of anization to determine the visual representation plex,
high-dimensional datasets. anizing systems g