Learning polynomial networks for classification of clinical electroencephalograms

Abstract

We describe a polynomial network technique developed for learning to classify clinical electroencephalograms (EEGs) presented by noisy features. Using an evolutionary strategy implemented within group method of data handling, we learn classification models which are comprehensively described by sets of short-term polynomials. The polynomial models were learnt to classify the EEGs recorded from Alzheimer and healthy patients and recognize the EEG artifacts. Comparing the performances of our technique and some machine learning methods we conclude that our technique can learn well-suited polynomial models which experts can find easy-to-understand.

Feature Extraction with GMDH-Type Neural Networks for EEG-Based Person Identification

“…28 Within the Deep Learning framework, neurons at new layers of neural networks are generated with new features which are capable of improving the ability to generalise. [28][29][30] The advantages of the GMDH approach have been shown in our early research, 31,32 in which a new algorithm developed for learning GMDH-type ANN has outperformed the conventional ANN trained with the back-propagation method. The methods have been compared in terms of the accuracy of classification of clinical EEG.…”

Financial distress prediction in banks using Group Method of Data Handling neural network, counter propagation neural network and fuzzy ARTMAP

“…Other applications include breast cancer diagnosis [17], classifying different soil samples [55], classifying bio-signal data [20], image processing [45], character recognition [9], classifying radar pulses in autonomous electronic support measure systems [21], gas identification using a microhotplate gas sensor [34], forecasting electric load [35], analyzing the stability of electric energy systems [18], applications of VLSI circuits [58], and wireless electronic nose system for analyzing gas mixtures [15]. GMDH was applied in complex modeling system of a mandarin tree water usage [30], feature selection [1], fault diagnosis [36], classifying clinical electroencephalograms (EEGs) [52], and forecasting energy demand [56]. CPNN was applied in fault detection in a non-isothermal continuous stirred tank reactor (CSTR) [42], star identification [40], online handwritten character recognition [66], automatic incident detection system for freeways [57].…”

Fault diagnosis and condition surveillance for plant rotating machinery using partially-linearized neural network

“…Although fault diagnosis of rolling bearings is often artificially carried out using time or frequency analysis of vibration signals, there is a need for a reliable, fast automated diagnosis method thereof. Neural Networks (NN) have potential applications in automated detection and diagnosis of machine failures (Yang, Stronach, & MacConnell, 2003;Samanta, Al-Balushi, & Al-Araimi, 2006;Schetinin & Schult, 2006;Li, Chen, & Wu, 2006;Samanta & Al-Balushi, 2003;Alguindigue, Loskiewizc-Buczak, & Uhric, 1993;Tao, Li, & Fang, 2006;Saxena & Saad, 2007;Su & Chong, 2007). However a conventional NN cannot adequately reflect the possibility of ambiguous diagnosis problems, and will never converge, when the symptom parameters, input to the 1st layer of the NN, have the same values in different states (Bishop, 1995).…”

“…Furthermore, although many studies have investigated the use of neural networks for fault diagnosis and condition surveillance of machinery, most of these studies have dealt with the discrimination of fault types (Yang et al, 2003;Samanta et al, 2006;Schetinin & Schult, 2006;Li et al, 2006). In these cases, training data for the neural network must be measured in the normal state and in each abnormal state for the discrimination.…”