In this paper, we present a machine learning-based approach that leverages Long Short-Term Memory (LSTM) networks combined with sliding window technique for feature extraction, aimed at accurately predicting point defect percentages in semiconductor materials based on simulated X-ray Diffraction (XRD) data. The model was initially trained silicon-simulated XRD data ranging from 1 to 5%, enabling it predict 0 10% silicon and other materials, including AlAs, CdS, GaAs, Ge, ZnS. Through...

Although backpropagation is widely accepted as a training algorithm for artificial neural networks, researchers are always looking inspiration from the brain to find ways with potentially better performance. Forward-Forward new that more similar what occurs in brain, although there significant performance gap compared backpropagation. In algorithm, loss functions placed after each layer, and updating of layer done using two local forward passes one backward pass. its early stages has been...

Finger-Knuckle-Print (FKP) is an accurate and reliable biometric in compare to other hand-based biometrics like fingerprint because of the finger's dorsal region not exposed surfaces. In this paper, a simple end-to-end method based on Convolutional Neural Network (CNN) proposed for FKP recognition. The model composed only three convolutional layers two fully connected layers. number trainable parameters hereby has significantly reduced. Additionally, straightforward utilized data...

In this study, an illumination-tolerant face recognition algorithm is proposed. This work highlights the significance of matrix polar decomposition for illumination-invariant recognition. The proposed has two stages. first stage, authors reduce effect illumination changes by weakening discrete cosine transform coefficients block intensities using a new designed quantisation table. second unitary factor reconstructed image used as feature matrix. phase, novel indirect method measuring...

Abstract Objective. Object recognition and making a choice regarding the recognized object is pivotal for most animals. This process in brain contains information representation decision steps which both take different amount of times objects. While dynamics are usually ignored models, here we proposed fully spiking hierarchical model, explaining from to decision. Approach. Coupling deep neural network recurrent attractor based model beside using spike time dependent plasticity learning...

The study of handwritten words is tied to the development recognition methods be used in real-world applications involving words, such as texts, bank checks, and postal envelopes, among others. In this paper an approach for Farsi checks was proposed which legal amounts are aid results courtesy amounts. set some 40 specified divided into their sub- words. Some if-then rules extracted from validation validate recognized digits. These confirm, correct or reject digit. experimental reveal a rate...

This paper proposed a method based on Zernike moments to classify the various stages of Alzheimer's Disease(AD) from structural MRIs. The is benefited all three orthogonal directions MRIs i.e. Axial, Sagittal and Coronal images. Three back-propagation algorithms had been used train neural network with seven neurons in hidden layer reach best accuracy. We experimented this 232 OASIS database. 70 percent subjects for training other 30 was evaluate trained network. achieved accuracy 86.46...

Deep Learning (DL) has been recently utilized for image fusion applications. The aim of DL based multi-focus methods is to create the better decision map fusing input images compared with previous traditional in spatial and transform domains. Hence, Convolution Neural Networks (CNN) Fully (FCN) that were used recent have unsuitable initial segmented map, their architectures a large number parameters which need be updated during training process. This paper proposed simple method inspired by...