Meta-cognitive RBF Network and its Projection Based Learning algorithm for classification problems

Liver disorder diagnosis using linear, nonlinear and decision tree classification algorithms

“…The structure was designed with ten inputs, one hidden and one output layer. Initially multiple number of nodes (4,6,8,16,17,21) were tested in the hidden layer. The range of epochs was set from 10 to 1000.…”

“…These algorithms include artificial neural network (ANN), artificial immune system (AIS), case-based reasoning (CBR), classification and regression tree (CART), C4.5 and C5.0 decision trees, fuzzy logic (FL), rule-based reasoning (RBR) and support vector machines (SVMs) [1]. ANNs have been used by Hamamoto et al (1995) to predict early prognosis of hepatectomised patient with hepatocellular carcinoma [6], by Hayashi et al (2000) to diagnose hepatobiliary disorders [7], by Ozyilmaz and Yildirim (2003) to diagnose hepatitis disease [8], by Lee et al (2005) to classify liver cyst, hepatoma and cavernous haemangioma [9], by Yahagi (2005) to diagnose types of cirrhosis [10], by Azaid et al (2006) to classify fatty liver, liver cirrhosis and liver cancer [11], by Revett et al (2006) to perform mining of primary biliary cirrhosis [12] Babu and Suresh (2013) to classify liver disorder as sick and healthy [14][15][16][17], by Dong et al (2008) to calculate optimal value of cost parameter in order to minimize classification error [18], by Rouhani and Haghighi (2009), Ansari et al (2011) and Sartakhti et al (2015) to diagnose hepatitis disease [19][20][21], by Uttreshwar and Ghatol (2009) to specifically diagnose hepatitis B [22], by Bucak and Baki (2010) to classify liver disorders as hepatitis B, hepatitis C and cirrhosis [2], by Hashem et al (2010) to predict hepatic fibrosis extent in patients with HCV [23], by Revesz and Triplet (2010) to diagnosis primary biliary cirrhosis …”

Computer Aided Diagnosis of ADHD Using Brain Magnetic Resonance Images

“…Of these various models, the model proposed by Nelson and Narens [19] is the most comprehensive and imitable in machine learning. Several machine learning algorithms have been developed based on this Nelson and Narens model of human meta-cognition for real-valued neural networks [20,12], complex-valued neural networks [21][22][23] and neuro-fuzzy inference systems [24,25]. It can be seen from these studies that a meta-cognitive network with self-regulated learning outperform the networks without meta-cognition.…”

“…First, the ROI of these regions are defined using the Wake Forest University Pick-atlas [11]. The voxels of the ROI thus obtained are then used as input features to train a Projection Based Learning algorithm of a Meta-cognitive Radial Basis Function Network (PBL-McRBFN) classifier [12] to perform ADHD diagnosis. Based on the diagnostic study, the subjects are classified either as Typically Developing Controls (TDC) or ADHD.…”

“…As the ADHD 200 competition data set is a real-valued data set, we use the real-valued meta-cognitive neural network in our work. Of the real-valued meta-cognitive neural networks, the projection based learning algorithm of the meta-cognitive radial basis function network [12] performs classification better than the meta-cognitive neural network [20] due to its ability to store past knowledge and to use the knowledge at a later stage of learning. Hence, in this work, we employ the projection based learning algorithm of the meta-cognitive radial basis function network [12] for the diagnosis of ADHD and identification of its subtype.…”