Wyniki wyszukiwania - BazTech

1

DCNN oscillator design, implementation, and performance evaluation

Salih Thair A., Mohammed Emad A., Abdal Yasir M.

Przegląd Elektrotechniczny

|

2023

|

R. 99, nr 4

84--88

EN

FPAAs technology are the ideal solution for creating an analog system. FPAA describes and implements the architecture of a simple CNN model used to build a delayed cellular neural network (DCNN) oscillator in this letter. Matlab simulation was carried out in order to analyze the proposed system. The attraction spectrum with different initial conditions, as well as Lyapunov's motives, were used to investigate the fundamental characteristics of the proposed system. The effect of noise on the proposed system was investigated. The promising results obtained encourage the application of the proposed model in secure communication systems.

PL

Technologia FPAA jest idealnym rozwiązaniem do stworzenia systemu analogowego. W tym liście FPAA opisuje i implementuje architekturę prostego modelu CNN używanego do budowy oscylatora opóźnionej komórkowej sieci neuronowej (DCNN). W celu analizy proponowanego systemu przeprowadzono symulację w Matlabie. Widmo przyciągania z różnymi warunkami początkowymi, a także motywy Lapunowa zostały wykorzystane do zbadania podstawowych cech proponowanego systemu. Zbadano wpływ hałasu na proponowany system. Uzyskane obiecujące wyniki zachęcają do zastosowania proponowanego modelu w bezpiecznych systemach komunikacyjnych.

2

Deep convolutional neural network for chronic kidney disease prediction using ultrasound imaging

Patil Smitha, Choudhary Savita

Bio-Algorithms and Med-Systems

|

2021

|

Vol. 17, no. 2

137--163

EN

Objectives: Chronic kidney disease (CKD) is a common disease and it is related to a higher risk of cardiovascular disease and end-stage renal disease that can be prevented by the earlier recognition and diagnosis of individuals at risk. Even though risk factors for CKD have been recognized, the effectiveness of CKD risk classification via prediction models remains uncertain. This paper intends to introduce a new predictive model for CKD using US image. Methods: The proposed model includes three main phases “(1) preprocessing, (2) feature extraction, (3) and classification.” In the first phase, the input image is subjected to preprocessing, which deploys image inpainting and median filtering processes. After preprocessing, feature extraction takes place under four cases; (a) texture analysis to detect the characteristics of texture, (b) proposed highlevel feature enabled local binary pattern (LBP) extraction, (c) area based feature extraction, and (d) mean intensity based feature extraction. These extracted features are then subjected for classification, where “optimized deep convolutional neural network (DCNN)” is used. In order to make the prediction more accurate, the weight and the activation function of DCNN are optimally chosen by a new hybrid model termed as diversity maintained hybrid whale moth flame optimization (DM-HWM) model. Results: The accuracy of adopted model at 40th training percentage was 44.72, 11.02, 5.59, 3.92, 3.92, 3.57, 2.59, 1.71, 1.68, and 0.42% superior to traditional artificial neural networks (ANN), support vector machine (SVM), NB, J48, NBtree, LR, composite hypercube on iterated random projection (CHIRP), CNN, moth flame optimization (MFO), and whale optimization algorithm (WOA) models. Conclusions: Finally, the superiority of the adopted scheme is validated over other conventional models in terms of various measures.

3

Comparative Experimental Investigation and Application of Five Classic Pre-Trained Deep Convolutional Neural Networks via Transfer Learning for Diagnosis of Breast Cancer

Nogay Hidir Selcuk, Akinci Tahir Cetin Akinci, Yilmaz Musa

Advances in Science and Technology. Research Journal

|

2021

|

Vol. 15, no 3

1--8

EN

In this study, for the diagnosis and classification of breast cancer, we used and applied five classical pre-trained deep convolutional neural network models (DCNN) which have proven successful many times in different fields (ResNet-18, AlexNet, GoogleNet and SuffleNet). To make pre-trained DCNN models suitable for the purpose of our study, we updated some layers according to the new situation by using the transfer learning technique. We did not change the weights of all layers used in these five pre-trained DCNN models. Instead, we just gave new weights to the new layers so that new layers adapt faster to emerging new DCNN models. With these five pre-trained DCNN models, we have realized a quadruple classification as "cancer", "normal", "actionable" and "benign", and a binary classification as "actionable + cancer" and "normal + benign". With these two separate classification and diagnosis studies, we have carried out comparative experimental examination and analysis of pre-trained DCNN models for breast cancer diagnosis. In the study, it was concluded that successful results can be achieved with pre-trained DCNN models without extra time-consuming procedures such as feature extraction, and DCNN can perform quite successfully in cancer diagnosis and image comment.