Title :
Right whale detection using artificial neural network and principal component analysis
Author :
Pylypenko, Kostiantyn
Author_Institution :
Dept. of Acoust. & Acoustoelectronics, Nat. Tech. Univ. of Ukraine “KPI”, Kiev, Ukraine
Abstract :
In this paper applying PCA algorithm in order to preprocess data in North whale detection problem was proposed. It is shown that reducing number of features obtained from a spectrogram can prevent overfitting and increase performance of the classifier. Proposed classifier was based on a tree layer Neural Network with 50, 100 and 400 input neurons. Proposed technique achieved an F1 score of 90%.
Keywords :
neural nets; principal component analysis; signal classification; signal detection; trees (mathematics); zoology; PCA algorithm; artificial neural network; classifier; north whale detection problem; principal component analysis; right whale detection; spectrogram; tree layer neural network; Acoustics; Artificial neural networks; Conferences; Image reconstruction; Principal component analysis; Spectrogram; Whales; Artificial Neural Network; Principal Component Analysis; spectrogram; whale detection;
Conference_Titel :
Electronics and Nanotechnology (ELNANO), 2015 IEEE 35th International Conference on
Conference_Location :
Kiev
DOI :
10.1109/ELNANO.2015.7146912
