Data Mining and Knowledge Engineering

R. Bhuvana ¹, S. Purushothaman ², R. Rajeswari ³

Affiliations
1 Department of Computer Science, Vels University, Chennai, IN
2 PET Engineering College, Vallioor-627117, IN
3 Mother Teresa Womens University, Kodaikanal, IN

Abstract

Depression is a serious and widespread public health challenge. This paper propose neural network algorithm for faster learning of psychological depression data. Implementation of neural networks methods for depression data mining using back propagation algorithm (BPA) and Echo state neural network (ESNN) are presented. Experimental data were collected with 21 input variables and one output for working with artificial neural network (ANN). Using the data collected, the training patterns and test patterns are obtained. The input patterns are pre-processed and presented to the input layer of ANN In order to find the optimum number of nodes required in the hidden layer of an ANN, a method has been proposed, based on the change in the mean squared error dynamically, during the successive sets of iterations. The output of BPA is given as input to ESNN. The network trained with transformed vectors is seen to require the least computational effort. The work proves to be an efficient system for diagnosis of depression.

Keywords

Hamilton Rating Scale (HRS) Depression Data, BPA, ESNN.

Full Text

     

B. Shaji ¹, S. Purushothaman ², R. Rajeswari ³

Affiliations
1 VELS University, Pallavaram, Chennai-600117, IN
2 PET Engineering College, Vallioor-627117, IN
3 Mother Teresa Women's University, Kodaikanal-624102, IN

Abstract

Wavelet decomposition has been applied to mammogram image to obtain four different coefficients for 5 levels of decompositions. The coefficients are: low frequency coefficients (A), vertical high frequency coefficients (V), horizontal high frequency coefficients (H), and diagonal high frequency coefficients (D). The features of the mammography image are obtained using the wavelet transform selecting the different levels of decompositions. The proposed method presents a new classification approach to microcalcification (MC) detection in mammograms using wavelet and back propagation algorithm (BPA) Neural Network. These features obtained from wavelet are representation of MC as well as other information of the image. Daubauchi wavelet has been used to decompose image to 5 levels. Statistical features are extracted from the wavelet coefficients. Training the BPA with features and testing the BPA to identify the presence of MC has been done. The percentage identification is above 96.2%. The performance of the proposed method based on the quality of the mammogram image.

Keywords

Mammogram, Microcalcification, Wavelet Transform, Back Propagation Neural Network.

Full Text

     

Y. Kalpana ¹, S. Purushothaman ², R. Rajeswari ³

Affiliations
1 VELS University, Pallavaram, Chennai-600117, IN
2 PET Engineering College, Vallioor-627117, IN
3 Mother Teresa Women's University, Kodaikanal-624102, IN

Abstract

Intrusion detection is the art of detecting computer abuse and any attempt to break into networks. As a field of research, it must continuously change and evolve to keep up with new types of attacks or adversaries and the ever-changing environment of the Internet. To make networks more secure, intrusion-detection systems (IDS) aims to recognize attacks. Artificial neural networks (ANN) based IDS were implemented and tested. The goal for using ANNs for intrusion detection is to generalize from incomplete data and able to classify data as being normal or intrusive. An ANN consists of a collection of processing elements that are highly interconnected. Given a set of inputs and a set of desired outputs, the transformation from input to output is determined by the weights associated with the inter-connections among processing elements. By modifying these interconnections, the network adapts to desired outputs. The ability of high tolerance for learning-by-example makes neural networks flexible and powerful in IDS. This paper has implemented Echo state neural network and Radial basis function applied to intrusion detection. The scope of the work includes using the available KDD database.

Keywords

Radial Basis Function (RBF) Networks Echo State Neural Networks (ESNN), KDD Features, Intrusion Detection.

Full Text