An efficient neural network architecture for recognition of spatial pattern invariants
The authors describe a neural architecture for efficient recognition of invariant features placed arbitrarily in patterns of data. The architecture provides versatility in invariant selection with minimal computation and storage requirements. Operating in dumb mode, the architecture, called the big,...
Gespeichert in:
Hauptverfasser: | , |
---|---|
Format: | Tagungsbericht |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | The authors describe a neural architecture for efficient recognition of invariant features placed arbitrarily in patterns of data. The architecture provides versatility in invariant selection with minimal computation and storage requirements. Operating in dumb mode, the architecture, called the big, dumb mass detector (BDMD), autonomously extracts fixed-size subsets of pattern components based upon total activation in the receptive fields of detector nodes. The BDMD sends these subset patterns to a pattern recognition neural network (PR network, an ART system) whose input field has exactly as many nodes as the fixed-size subsets. Operating in smart mode, the BDMD can be driven by another part of the recognition system to selectively scan an input pattern. Among the capabilities of the BDMD is its ability to zoom in on a pattern feature of particular interest. Only the relatively few connections and associated computations required for recognizing single features need to be implemented in the PR network.< > |
---|---|
DOI: | 10.1109/IJCNN.1992.227340 |