The relationship between immunological memory and a class of associative memories known as sparse distributed memories (SDM) is well known. This paper proposes a new model for clustering non-stationary data based on a combination of salient features from the two metaphors. The resulting system embodies the important principles of both types of memory; it is self-organising, robust, scalable, dynamic and can perform anomaly detection. The model is rst applied to clustering static datasets, and is shown to outperform two other systems based on immunological principles. It is then applied to clustering non-stationary data-sets with promising results.

The vertebrate immune system is a robust and powerful information process system that demonstrates features such as distributed control, parallel processing and adaptation and learning via experience. Artificial Immune Systems (AIS) are machine-learning algorithms that embody some of the principles and attempt to take advantages of the benefits of natural immune systems for use in tackling complex problem domains. The Immunos-81 is an AIS technique designed for classification problem domains. It is a technique that has been vaguely described and mentioned in the field of AIS research, though has not been investigated in depth nor has the algorithm or its results been reproduced. This work rigorously analyses the proposed classification system and describes both its biological inspiration and its computational implementation in detail. Two implementations are provided, that reproduces the general themes of the approach and show similar results. Finally, the general themes of the system are integrated with elements from clonal selection inspired algorithms. A new classification algorithm is designed, implemented and tested called Immunos-99 that exhibits desirable