Detecting rare events with lotto-type competitive learning

This paper highlights the difficulty of detecting small and rare clusters. In theory it is possible for individual neurons, in lotto-type competitive learning algorithms, to follow the source density function. We note that in experiment it is very difficult to locate these small clusters, especially if the prototype set is limited. Two methods are proposed, as exploratory tools, to locate these clusters. The first method is to train the network iteratively with the same prototype set. This simple method enables us to locate these small clusters, albeit with the possibility of over-training. The second method is to deploy supervisory agent(s) to track the trajectory of individual neurons.