An analysis of self-organization process for data classification in multisensor systems

Slawomir Przylucki; Waldemar Wojcik; Konrad Plachecki; Tomasz Golec

doi:10.1117/12.517138

Translator Disclaimer

22 September 2003 An analysis of self-organization process for data classification in multisensor systems

Slawomir Przylucki, Waldemar Wojcik, Konrad Plachecki, Tomasz Golec

Author Affiliations +

Proceedings Volume 5124, Optoelectronic and Electronic Sensors V; (2003) https://doi.org/10.1117/12.517138
Event: Optoelectronic and Electronic Sensors V, 2002, Rzeszow, Poland

Abstract

In this paper we present the idea of the optoelectronic measurement system for monitoring the industrial gas pollutants. The system consists of an optical detection system, an optical fiber as a data transmission link, a spectrometer with linear diode array and a neural network unit for real-time spectral data processing. We paid main attention to the neural network structure and its properties for gas recognition and gas concentration estimation task. The article presents the new classification algorithm based on the selforganizing artificial neural network. The algorithm comes from the kohonen selforganizing neural net idea. It introduces the groups of winners and that is why, we call it Multi-Winners Selforganizing Kohonen Map - MWSOM. The behavior and fundamental featured of that classifier are presented and verified by comparison to other 'classical' classification algorithms.

Citation Download Citation

Slawomir Przylucki, Waldemar Wojcik, Konrad Plachecki, and Tomasz Golec "An analysis of self-organization process for data classification in multisensor systems", Proc. SPIE 5124, Optoelectronic and Electronic Sensors V, (22 September 2003); https://doi.org/10.1117/12.517138

ACCESS THE FULL ARTICLE