Performance Comparison of Multilayer Neural Networks for Sleep Snoring Detection
Tan Loc Nguyen, Yonggwan Won · 2014
Sleep snoring is becoming a big social issue for long term healthcare, because it is related to other critical diseases. Recently, a novel Multilayer Perceptron neural network (MLP) which has the first hidden layer of correlational filter operation, named as f-MLP, was proposed. It demonstrated a superior classification performance for the pattern sets in which the frequency information is the dominant feature for classification. In this paper, we report the performance comparison of this f-MLP with the ordinary MLP. As a result, the f-MLP achieved an average over 95% classification rate for the test patterns, which is superior to the ordinary multilayer neural network that demonstrates an average about 84%.