HYDROACOUSTICS
ANNUAL JOURNAL
START NEW VOL 20 SEARCH STATISTICS PAS - GDANSK DIVISION

Wieloczęstotliwościowe wieloetapowe klasyfikatory neuronowo-rozmyte do rozpoznawania typu dna morskiego

pp. 39-44, vol. 3, 2000

T. V. Dung
Politechnika Gdańska, Katedra Telemonitoringu Środowiska, Gdańsk, Poland

Andrzej Stepnowski
Politechnika Gdańska, Katedra Telemonitoringu Środowiska, Gdańsk, Poland

Key words:

Abstract: A hybrid multistage neuro-fuzzy classifiers were developed for sea-bottom recognition from acoustic echoes. A multistage fuzzy neural network was implemented and tested on the data collected on two echosounder's frequencies. Two structures termed as incremental fuzz neural network (IFNN) and aggregated fuzzy neural network (AFNN), were analysed. In IFNN, an approximate decision is undertaken firstly, based only on the one set of input variables. The decision is then fine-tuned by considering more factors in following stages until the final decision, assigning the output class, is undertaken. In AFNN, the input variables are divided into M subsets, where each of them is fed to one sub-stage. The final output is derived by the reasoning with alt intermediate variables, which work as the outputs of substages in the preceding stage. The proposed structures improve the generalisation ability of the system and reduces requirements on computation power and memory.

Download: Fulltext PDF, BibTeX

© Polish Acoustical Society - Gdansk Department, Polish Academy of Sciences. This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported. (CC BY-NC-SA 3.0)