StreamSoNG: A Soft Streaming Classification Approach

Examining most streaming clustering algorithms leads to the understanding that they are actually incremental classification models. They model existing and newly discovered structures via summary information we call footprints. Incoming data is normally assigned a crisp label (into one of structures) structure's footprint incrementally updated. There no reason these assignments need be crisp. In this paper, propose new algorithm uses Neural Gas prototypes as footprints produces possibilistic vector (of typicalities) for each incoming vector. These typicalities generated by modified k-nearest neighbor algorithm. The approach tested on synthetic real image datasets. We compare our three other classifiers based Adaptive Random Forest, Very Fast Decision Rules, DenStream with excellent results.