Cloud-based architecture for face identification with deep learning using convolutional neural network

Abstract

The use of a face as a biometric to identify a person in order to keep the system safe from an unauthorized person has advantages over other biometric characteristics. The face as a biometric has more structure and a wider area than other biometrics, while can be retrieved in a non-invasive manner. We proposed a cloud-based architecture for face identification with deep learning using convolutional neural network. Face identification in this study used a cloud-based engine with four stages, namely face detection with histogram of oriented gradients (HOG), image enhancement, feature extraction using convolutional neural network, and classification using k-nearest neighbor (KNN), SVM, as well as random forest algorithm. This study conducted a classification experiment with cloud-based architecture using three different datasets, namely Faces94, Faces96 and University of Manchester Institute of Science and Technology (UMIST) face dataset. The results from this study are with the proposed cloud-based architecture, the best accuracy is obtained by KNN algorithm with an accuracy of 99% on Faces94 dataset, 99% accuracy on Faces96 dataset, 97% on UMIST face dataset, and performance of the three algorithms decreased in UMIST face dataset with facial variations from various angles from left to right profile.

Distributed memory of neural networks and the problem of the intelligence`s essence

“…The main idea of constructing such codes is based on the use of redundant information. So, to transmit one of 2 4 code ISSN: 2302-9285  Distributed memory of neural networks and the problem of the intelligence`s … (Ibragim E. Suleimenov) 513 combinations, the binary (7,4) Hamming code uses 7 binary variables [24]. To simplify, one can say that 7 bits are used in order to convey the same amount of information that could be transmitted (with the deliberate absence of errors) using only 4 bits.…”

“…To simplify, one can say that 7 bits are used in order to convey the same amount of information that could be transmitted (with the deliberate absence of errors) using only 4 bits. From the point of view of set theory as shown in Figure 3, the use of error correction code can be interpreted as; the set A of all possible code combinations (in the example of the binary (7,4) Hamming code considered above-2 7 ) is divided into subsets A i , the number of which is equal to the number of code combinations r, treated as a code with an error corrected (in this example, 2 4 ). Any a ∈ A i is associated with a code combination with an absent error from the set B.…”

“…It is often emphasized that memory of artificial neural networks (ANN) is distributed in current literature [1], [2]. It is pertinent to emphasize that this conclusion was actually made on an empirical basis, as well as a significant part of other conclusions that in one way or another relate to the properties of ANNs [3], [4]. Until now, there are no algorithms that would allow calculating the weight coefficients of a specific ANNs, based on the requirements connected with the solving problem.…”

An experimental study of tomato viral leaf diseases detection using machine learning classification techniques

“…Major states in India produce tomatoes [2] Figure 2. Sample of major diseases of tomato in India (NHB, India database) Several studies were done earlier that show the capability of technology using algorithms of machine learning to identify objects in various sectors like retail, human behavior, face recognition [3], facial expressions [4], handwriting recognition [5], intrusion detection [6], movie recommendation [7], and food segmentation [8]. The health sector is also using it to identify/detect/predict various diseases like diabetes prediction, cancer detection [9], [10], heart disease, skin problems, Parkison's disease identification [11], COVID-19 from chest X-rays [12], and many more.…”