BP Neural Network (BPNN)

Machine Learning can refer to a different and various algorithms, based on artificial intelligence, that are able to recognize data patterns through continuous and repeated learning techniques, and we do not have to assume any prior data distribution. In this way, artificial neural networks can provide an effective technology and methodology. Artificial neural networks need a rule, delta rule, in order to learn from the supplied data. In this paper a generalized delta rule with a new error function is presented. Keywords: neural networks, delta rule, error function, cost function.

Springer Series in the Data Sciences focuses primarily on monographs and graduate level textbooks. The target audience includes students and researchers working in and across the fields of mathematics, theoretical computer science, and statistics. Data Analysis and Interpretation is a broad field encompassing some of the fastest-growing subjects in interdisciplinary statistics, mathematics and computer science. It encompasses a process of inspecting, cleaning, transforming, and modeling data with the goal of discovering useful information, suggesting conclusions, and supporting decision making. Data analysis has multiple facets and approaches, including diverse techniques under a variety of names, in different business, science, and social science domains. Springer Series in the Data Sciences addresses the needs of a broad spectrum of scientists and students who are utilizing quantitative methods in their daily research. The series is broad but structured, including topics within all core areas of the data sciences. The breadth of the series reflects the variation of scholarly projects currently underway in the field of machine learning.

As the global energy crisis and environmental pollution intensify, energy conservation and emission reduction have become essential objectives for the development of smart campuses at universities and colleges worldwide. Effective management of energy consumption in these environments relies on accurate power consumption prediction models. This paper investigates the prediction of power consumption for smart campuses through a meteorological data-driven model. By analyzing key meteorological factors, such as temperature, humidity, wind speed, and solar radiation, which influence electricity consumption on campus, we propose a prediction model using a genetic algorithm-backpropagation (GA-BP) neural network combined with sample entropy value and assignment methods. This model utilizes actual data on campus electricity consumption, augmented with multi-source meteorological data for training and validation purposes. Experimental results indicate that the model achieves high prediction accuracy across various meteorological data-driven conditions, thus enhancing the optimization of energy consumption management on campuses. The findings of this research offer theoretical support for energy-saving strategies in smart campuses and serve as a valuable reference for predicting energy consumption in similar public buildings.

Software testing is an important discipline, and consumes significant amount of effort. A proper strategy is required to design and generate test cases systematically and effectively. In this paper automated software test case generation with Radial Basis Function Neural Network (RBFNN) has been proposed and empirically validated with the help of a case study and compared with other techniques of soft computing. Experimental results show that RBFNN is one of the best technique for automated test case generation

Artificial neural networks can produce solutions for samples that have not been addressed before, by generalizing the relationship between inputs and outputs related to a problem from existing samples. In the studies conducted in this area, it has been pointed out that the artificial neural network models are similar to some statistical methods, while the working principles of some of them are almost the same. Because of the similarities between these two disciplines, they need to be compared with each other to show how important one is in the development of the other. In this study, it is aimed to investigate the usability and effectiveness of artificial neural network models in binary classification problems. For this purpose, firstly, artificial neural network models are briefly mentioned. Then, the similarities between them and some statistical methods were taken into consideration. In the application phase, logistic regression analysis, which is frequently used in binary classification problems, and artificial neural network models were applied on a data set obtained from patients who consulted the Internal

A single layer perceptron is a simplest form of neural network. This type of neural network is used for pattern classifications that are linearly separable. Single layer perceptron consists of one input layer with one or many input units and one output layer with one or many output units. The neurons are connected to each other by weights and bias (Gkanogiannis and Kalamboukis (2009, 2010)).

In this paper the artificial neural network training algorithm is implemented in MATLAB language. This implementation is focused on the network parameters in order to get the optimal architecture of the network that means (the optimal neural network is the network that can reach the goals in minimum number of training iterations and minimum time of training). Many examples were tested and it was shown that using one hidden layer with number of neuron equal to the square of the number of inputs will lead to optimal neural network by mean of reducing the number of training stages (number of training iterations) and thus the processing time needed to train the network. Our goal in this paper is reaching the optimal solution by reaching minimum training time and minimum number of training iterations.

Artificial neural networks can produce solutions for samples that have not been addressed before, by generalizing the relationship between inputs and outputs related to a problem from existing samples. In the studies conducted in this area, it has been pointed out that the artificial neural network models are similar to some statistical methods, while the working principles of some of them are almost the same. Because of the similarities between these two disciplines, they need to be compared with each other to show how important one is in the development of the other. In this study, it is aimed to investigate the usability and effectiveness of artificial neural network models in binary classification problems. For this purpose, firstly, artificial neural network models are briefly mentioned. Then, the similarities between them and some statistical methods were taken into consideration. In the application phase, logistic regression analysis, which is frequently used in binary classification problems, and artificial neural network models were applied on a data set obtained from patients who consulted the Internal Medicine polyclinic of Osmangazi University Health, Application and Research Hospital, and the results were compared. According to the results obtained, it was observed that artificial neural network models gave better results than logistic regression analysis in binary classification problems.

Developments in deep learning with ANNs (Artificial Neural Networks) are paving the way for revolutionizing all application areas, especially related to non-linear regression and classification problems of predictive modelling and forecasting. Although their explainability is more complicated and challenging, deep neural networks are preferred over conventional machine learning methods for high accuracy in non-linear and complex problems. However, machine learning and data science practitioners often use ANN like a black-box. The present article concisely overviews the mathematics and computations involved in simple feed-forward neural networks (FNNs) or multilayer perceptrons (MLPs). The purpose is to spot light on what deep neural networks' learning (or training) is and how it works. The article includes simplified derivations of the expressions for the main workhorse of neural networks (the backpropagation) and an example to explain how it works with graphical insights. An algorithm for a basic ANN application is presented in both component-form and matrix-form, together with a detailed note on the relevant data structures, to elaborate the scheme comprehensively. Python implementation of the basic algorithm is presented, and its performance results are compared with those produced using the TensorFlow library functions that implement the neural networks. The article discusses various techniques to improve the generalization capability of neural networks and how to address various training challenges. Finally, some well-established optimization approaches based on the Gradient Descent method are also discussed. The article may serve as a comprehensive premiere for a sound understanding of deep learning for undergraduate and graduate students before indulging in the relevant industry practices so that they can step into sustainable progress in the field.

This study employed an artificial neural network for population prediction (ANNPP) that handles incomplete and inconsistent nature of data usually experienced in the use of mathematical and demographic models while carrying out population prediction. ANNPP uses the three demographic variables of fertility, mortality and migration which are the major dynamics of population change as the input data. The datasets were divided into train, validation and test data. The train data was presented to the supervised artificial network to approximate some known twelve target values of population growth rates. The method was also used to simulate both the validation and the test datasets as case data on the consistency of results obtained from the training session via the train data. From the sixteen different topologies tested on the basis of the mean square errors (MSE), standard deviation (STDEV) and epochs; topology 19-9-1 performed best than the rest. A comparison between the predictions based on the ANNPP derived growth rates and The cohort component method of population prediction (CCMPP) was compared. The results showed that ANNPP percentage accuracies ranged between 81.02 and 99.15% while that of CCMPP percentage accuracies ranged between 64.55 and 86.43%. These results showed that artificial neural network model performed better than the demographic model.

This study employed an artificial neural network for population prediction (ANNPP) that handles incomplete and inconsistent nature of data usually experienced in the use of mathematical and demographic models while carrying out population prediction. ANNPP uses the three demographic variables of fertility, mortality and migration which are the major dynamics of population change as the input data. The datasets were divided into train, validation and test data. The train data was presented to the supervised artificial network to approximate some known twelve target values of population growth rates. The method was also used to simulate both the validation and the test datasets as case data on the consistency of results obtained from the training session via the train data. From the sixteen different topologies tested on the basis of the mean square errors (MSE), standard deviation (STDEV) and epochs; topology 19-9-1 performed best than the rest. A comparison between the predictions based on the ANNPP derived growth rates and The cohort component method of population prediction (CCMPP) was compared. The results showed that ANNPP percentage accuracies ranged between 81.02 and 99.15% while that of CCMPP percentage accuracies ranged between 64.55 and 86.43%. These results showed that artificial neural network model performed better than the demographic model.

Eixo 3. Tecnologias, mídias e processos inovadores na EaD 1 O sentido da mudança é distinto das transformações e está ligado as reformas na educação que mudam apenas superficialmente as práticas pedagógicas. Em geral, são de cima para baixo.

O presente trabalho busca compreender como os cursos de Pedagogia, na modalidade a distancia, abordam a literatura infantil, com foco na formação e desenvolvimento do Ser mulher. Para tal, analisou-se a grade curricular da modalidade a distancia da Universidade Federal de São Carlos (UFSCar) e do Instituto Federal de Rondônia (IFRO), fazendo um comparativo com a grade curricular da Universidade Federal de Goiás (UFG) que dispõe do curso de Pedagogia presencial, apresentando assim a urgência da importância da inserção de disciplinas que abordem assuntos relacionados a gênero em seu componente curricular, para que apresente um ensino democrático, consciente e engajado nas duas modalidades.

In Ethiopian history, agriculture has been the backbone of the economy. This agricultural activity remain undeveloped due to different factors. Most of the activities are done with a lack of modern technology. Currently, seed classification is done based on knowledge of human being. The current seed classification analysis is inefficient and has no validation mechanism. In this research, we have made an effort to present a predictive model to predict seed classes using machine learning algorithms which results in high crop production. For the development, this research machine learning algorithm is used to learn from data which can be used to make predictions, to make real-world simulations, for pattern recognitions and classifications of the input data. An artificial neural network is used for modelling complex relationships between inputs and outputs or to find patterns in data. The objective of this thesis is to understand the machine learning algorithm using neural networks and ...

Chaotic systems are nonlinear systems that show sensitive dependence on initial conditions, and an immeasurably small change in initial value causes an immeasurably large change in the future state of the system. Besides, there is no randomness in chaotic systems and they have an order within themselves. Researchers use chaotic systems in many areas such as mixer systems that can make more homogeneous mixtures, encryption systems that can be used with high security, and Artificial Neural Networks (ANNs) by taking the advantage of the order in this disorder. Differential equations in which chaotic systems are expressed mathematically are solved by numerical solution methods such as Heun, Euler, ODE45, RK4, RK5-Butcher and Dormand-Prince in the literature. In this research, Feed Forward Neural Network (FFNN), Layer Recurrent Neural Network (LRNN) and Cascade Forward Backpropogation Neural Network (CFNN) structures were used to model the Rucklidge chaotic system by making use of the MATLAB R2021A and Neural Network (NN) Toolbox. By comparing the results of different activation functions used in the modeling, the ANN structure that can best model the Rucklidge chaotic system has been determined. The training of the compared ANNs was carried out with the values obtained from the Euler numerical solution method, which can get satisfactory and fast results.

Chaotic systems are nonlinear systems that show sensitive dependence on initial conditions, and an immeasurably small change in initial value causes an immeasurably large change in the future state of the system. Besides, there is no randomness in chaotic systems and they have an order within themselves. Researchers use chaotic systems in many areas such as mixer systems that can make more homogeneous mixtures, encryption systems that can be used with high security, and artificial neural networks by taking the advantage of the order in this disorder. Differential equations in which chaotic systems are expressed mathematically are solved by numerical solution methods such as Heun, Euler, ODE45, RK4, RK5-Butcher and Dormand-Prince in the literature. In this research, Feed Forward Neural Network (FFNN), Layer Recurrent Neural Network (LRNN) and Cascade Forward Backpropogation Neural Network (CFNN) structures were used to model the Rucklidge chaotic system by making use of the MATLAB R2...

When the feature space undergoes changes, owing to different operating and environmental conditions, multi-aspect classification is almost a necessity in order to maintain the performance of the pattern recognition system and improve robustness and reliability in decision making. This is an important issue being investigated in ANN research, in many cases, the problems can be solved more effectively by combining one or two other techniques rather than implementing ANN exclusively. New learning methods, especially multiple classifier systems, are now actively studied and applied in pattern recognition. So, the main goal of this paper is to propose two hybrid models and compare your performance in complex pattern recognition problem: speech recognition and biomedical diagnosis. This paper compare, the performance obtained with (1) Multi-network RBF/LVQ structure, we use involves Learning Vector Quantization (LVQ) as a competitive decision processor and Radial Basis Function (RBF) as a classifier. (2) Hybrid HMM/MLP model using a Multi Layer-Perceptron (MLP) to estimate the Hidden Markov Models (HMM) emission probabilities. For preclassification, the k-means clustering algorithm is proposed to obtain optimum information for the biomedical and speech training data for the proposed hybrid models.

The main goal of this paper is to compare the performance which can be achieved by two different hybrid approaches analyzing their applications' potentiality on real world paradigms (speech recognition and medical diagnosis). We compare the performance obtained with (1) Multinetwork RBF/LVQ structure, we use involves Learning Vector Quantization (LVQ) as a competitive decision processor and Radial Basis Function (RBF) neural models is used as classifier. (2) Hybrid HMM/MLP system using a Multi Layer Perceptron (MLP) to estimate the Hidden Markov Models (HMM) emission probabilities.

In this note we study the empirical pricing American options. The pricing American option is an optimal stopping problem, which can be derived from a backward recursion such that in each step of the recursion one needs conditional expectations. For empirical pricing, [Longstaff and Schwartz (2001)] suggested to replace the conditional expectations by regression function estimates. We survey the current literature and the main techniques of nonparametric regression estimates, and derive new empirical pricing algorithms.

Essa reflexão perpassa o ensino híbrido e seus desafios no trabalho docente no Ensino Superior como uma possibilidade para se pensar a realidade tanto de formação quanto de atuação docente perante o uso da TDICs no processo de ensino-aprendizagem. Em suma, essa temática é importante devido sua intensidade no processo educacional incentivada pelo âmbito econômico e sociopolítico. Outrossim, aqui, evidencia discussões sobre o trabalho docente diante de formações acadêmicas que, a priori, visam experiências pautadas no ensino presencial e pouco vínculo as ferramentas tecnológicas, uma vez que nem todas instituições e alunos têm acesso ao mundo virtual como uma forma de aprendizagem. Por fim, essa pesquisa pautada em um levantamento bibliográfico com abordagem qualitativa visa problematizar e analisar os desafios entrelaçados nas potencialidades desse período tecnológico na qual a sociedade está inserida.

Digital Elevation Models (DEMs) are commonly used for environment, engineering, and architecture-related studies. One of the most important factors for the accuracy of DEM generation is the process of spatial interpolation, which is used for estimating the height values of the grid cells. The use of machine learning methods, such as artificial neural networks for spatial interpolation, contributes to spatial interpolation with more accuracy. In this study, the performances of FBNN interpolation based on different parameters such as the number of hidden layers and neurons, epoch number, processing time, and training functions (gradient optimization algorithms) were compared, and the differences were evaluated statistically using an analysis of variance (ANOVA) test. This research offers significant insights into the optimization of neural network gradients, with a particular focus on spatial interpolation. The accuracy of the Levenberg–Marquardt training function was the best, wherea...

Problem statement: Most important problems of medical diagnosis. When there is a cerebrovascular accident attach the chances of a successful treatment depends essentially on the early diagnosis. In practice the part of medical errors while diagnosing a stroke type comes to 20-45% even for experienced doctors and the scope of methods of neurovisualization at stroke diagnosis are limited. Approach: In this research study, attempt was made to model the application of Artificial Neural Networks to the classification of patient Cerebrovascular Accident Attack. The Network for the consisted of a three-layer feed forward artificial neural network with back-propagation error method. Results: Data were collected from 100 records of patients at Federal Medical Centre Owo, Nigeria and the Artificial Neural Networks classifier was trained using gradient decent backward propagation algorithm with flexible sigmoid activation function at one hidden layer, with 16 inputs nodes representing stroke onset symptoms at the input layer, 10 nodes at the hidden layer and one node at the output layer representing the type of the attack. Conclusion: The learning Rate γ was set between 0.1 and 0.9 while the epoch set at 150. Initial weight set at Rand (-0.5 and 0.5). The simulation results showed that the model was capable of producing a reasonable forecasting accuracy in short.

This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright holders if permission to publish in this form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may rectify in any future reprint. Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers.

Este estudo busca investigar em que medida a oferta de cursos de graduação realizados na modalidade a distância contribuiu para o crescimento da graduação ofertada pela UFGD, no período de 2013 a 2018. Realizou-se levantamento do que foi previsto para a implantação do Programa REUNI e para o PDI da UFGD, pois ambos previam a expansão da oferta de vagas na graduação. Com base em dados coletados do Censo da Educação Superior, verificou-se o quadro de matriculados no Brasil, na Região Centro-Oeste e na UFGD. A partir da análise, conclui-se que a contribuição dos cursos EaD é significativa nesse processo de expansão, considerando principalmente que atende a um público que geralmente não possui condições para frequentar a graduação presencial.

The dynamics of supervised learning play a main role in deep learning, which takes place in the parameter space of a multilayer perceptron (MLP). We review the history of supervised stochastic gradient learning, focusing on its singular structure and natural gradient. The parameter space includes singular regions in which parameters are not identifiable. One of our results is a full exploration of the dynamical behaviors of stochastic gradient learning in an elementary singular network. The bad news is its pathological nature, in which part of the singular region becomes an attractor and another part a repulser at the same time, forming a Milnor attractor. A learning trajectory is attracted by the attractor region, staying in it for a long time, before it escapes the singular region through the repulser region. This is typical of plateau phenomena in learning. We demonstrate the strange topology of a singular region by introducing blow-down coordinates, which are useful for analyzin...

Este ensaio objetiva apresentar a modalidade de Educação a Distância-EaD a partir da Universidade Aberta do Brasil-UAB, discutindo, em específico, a sua história, o financiamento e os profissionais envolvidos. Como metodologia, parte-se de um estudo exploratório, retomando bases históricas e teóricas acerca do tema. A fundamentação teórica se assenta, de um lado, nos estudos de especialistas da área da Educação e das Tecnologias Digitais de Comunicação e Informação-TDCI e, de outro, na legislação brasileira desenvolvida a respeito dessa temática. Como resultado parcial, depreende-se que o Programa UAB impactou positivamente a Educação brasileira, sobretudo o oferecimento de cursos do Ensino Superior a distância.

O texto discute a inovação na Educação a Distância, com o objetivo de configurar o estatuto multidimensional da inovação em Educação. A pesquisa de natureza teórica parte do pressuposto de que as concepções de inovação precisam superar a centralidade de alguns sinônimos como mudança, reforma, renovação, evitando-se o caminho da adoção acrítica à inovação. A partir da questão central – inovamos em EaD? - analisa as propostas que se pautam pelo inovacionismo e inovação tecnológica ancoradas na obsolescência programada da educação e no discurso de que o sistema educacional está atrasado e defasado em relação a evolução tecnológica precisando por isso inovar o que pode ser superado configurando-se à inovação um caráter multidimensional.

Eixo 05: Educação híbrida: uma tendência na educação superior Resumo: As reflexões apresentadas neste artigo fazem parte do Projeto de Hibridização dos cursos de graduação de uma Faculdade privada de Minas Gerais. As concepções sobre aprendizagem vão orientar a elaboração dos projetos e planos que norteiam a IES. Objetivou-se refletir sobre as teorias da aprendizagem visando elaborar um modelo de hibridização que dialogue com a formação de um egresso autônomo capaz de aprender a aprender e interagir com a tecnologia a serviço da transformação da sociedade. Com uma metodologia de caráter qualitativo com viés exploratório descritivo, estabeleceu-se reflexões acerca da construção de propostas metodológicas de hibridização permitidas por lei para os cursos de graduação, buscando desenvolver um projeto inovador, moderno e centrado no estudante. Palavras-chave: Teorias da Aprendizagem. Gerações da EaD. Hibridização Introdução As reflexões apresentadas neste artigo fazem parte do Projeto de Hibridização dos componentes curriculares dos cursos de graduação presenciais de uma Faculdade privada da Região Central de Minas Gerais. No Brasil, a normatização da semipresencialidade desde a Portaria nº 4.059/2004, já permitiu a hibridização do ensino. Para Belloni (2011), Gomes (2005) e Keegan (1996) conceituar a Educação à Distância (EaD) não é tarefa das mais simples, pois é amplo e a princípio, cobre várias formas , níveis

This research paper proposes an improved feature reduction and classification technique to identify mild and severe dementia from brain MRI data. The manual interpretation of changes in brain volume based on visual examination by radiologist or a physician may lead to missing diagnosis when a large number of MRIs are analyzed. To avoid the human error, an automated intelligent classification system is proposed which caters the need for classification of brain MRI after identifying abnormal MRI volume, for the diagnosis of dementia. In this research work, advanced classification techniques using Support Vector Machines based on Particle Swarm Optimisation and Genetic algorithm are compared. Feature reduction by wavelets and PCA are analysed. From this analysis, it is observed that the proposed classification of SVM based PSO is found to be efficient than SVM trained with GA and wavelet based feature reduction technique yields better results than PCA.

Due to the complexity and non-linearity of most ecological problems, artificial neural networks (ANNs) have attracted attention from ecologists and environmental scientists in recent years. As these networks are increasingly being used in ecology for modeling, simulation, function approximation, prediction, classification and data mining, this unique and self-contained book will be the first comprehensive treatment of this subject, by providing readers with overall and in-depth knowledge on algorithms, programs, and applications of ANNs in ecology. Moreover, a new area of ecology, i.e., computational ecology, is proposed and its scopes and objectives are defined and discussed.

Computational Ecology consists of two parts: the first describes the methods and algorithms of ANNs, interpretability and mathematical generalization of neural networks, Matlab neural network toolkit, etc., while the second provides case studies of applications of ANNs in ecology, Matlab codes, and comparisons of ANNs with conventional methods. This publication will be a valuable reference for research scientists, university teachers, graduate students and high-level undergraduates in the areas of ecology, environmental sciences, and computational science.

Contents:
Artificial Neural Networks: Principles, Theories and Algorithms:
Feedforward Neural Networks
Linear Neural Networks
Radial Basis Function Neural Networks
BP Neural Network
Self-Organizing Neural Networks
Feedback Neural Networks
Design and Customization of Artificial Neural Networks
Learning Theory, Architecture Choice and Interpretability of Neural Networks
Mathematical Foundations of Artificial Neural Networks
Matlab Neural Network Toolkit
Applications of Artificial Neural Networks in Ecology:
Dynamic Modeling of Survivor Process
Simulation of Plant Growth Process
Simulation of Food Intake Dynamics
Species Richness Estimation and Sampling Data Documentation
Modeling Arthropod Abundance from Plant Composition of Grassland Community
Pattern Recognition and Classification of Ecosystems and Functional Groups
Modeling Spatial Distribution of Arthropods
Risk Assessment of Species Invasion and Establishment
Prediction of Surface Ozone
Modeling Dispersion and Distribution of Oxide and Nitrate Pollutants
Modeling Terrestrial Biomass

Readership: Research scientists, university teachers, graduate students and high-level undergraduates in the area of ecology, environmental sciences and computational science.

In lip reading, selection of features and classifier plays crucial roles. Goal of this work is to compare the common feature extraction modules and classifiers. Two well-known image transformed models, namely Discrete Cosine Transform (DCT) and Discrete Wavelet Transform (DWT) are studied. A competent feature extraction module cascaded with a robust classifier can result a novel automatic lip reading system. We have compared performance of Back Propagation Neural Network (BPNN) algorithm with that of K-Nearest Neighborhood (KNN) algorithm. Both being form class of artificial intelligence needs training. Hence we have also examined the computational complexity associated with the training phase of both classifiers. The CUAVE database is used for experimentation and performance comparison. It is observed that BPNN is a better classifier thatn KNN.