Implementation Of Artificial Neural Networks Research Articles

Experimental implementation of artificial neural network for cost effective and non-intrusive performance estimation of air conditioning systems

Owing to the high variability of operating conditions and the complexity of dynamic phenomena occurring within air conditioning cycles, the realistic performance estimation of these systems remains an open question in this field. This paper demonstrates the applicability of a cost-effective estimation method based on an artificial neural network exclusively using four refrigerant temperatures as the network input. The experimental datasets are collected from a reference experimental facility. The system is operated with variable cooling load, outdoor temperature, and indoor temperature settings, as representative of the actual operation. The artificial neural network structure was optimized by considering the effect of previous time step inputs, number of neurons, sampling time, and number of training data. The results reveal that the developed model can successfully estimate the cooling capacity of an air conditioning system during on–off, continuous unsteady, and steady operation, using four temperature inputs with relative averaged error below 5%.

Implementation of Artificial Neural Networks In Predicting Students ‘English Understanding Level Using The Backpropagation Method

English is a language that unites humans in communicating with others. The existence of language differences can make it difficult for people to understand each other in dialogue. Therefore, the role of English is very useful to unite human communication. In this case, researchers make research to predict the level of understanding of students in English. Predicting the level of understanding of students in English is needed to determine the level of ability or understanding of students in English so that students can further enhance student abilities. English is very necessary for students to support a bright future. In this study implements the Artificial Neural Network in conducting research and applying the bacpropagation method in it. To complete this study, researchers used several criteria, namely: Reading References, Hearing from the Environment, Practicing, Utilizing Technology. Of the four criteria using the backpropagation method is useful for training in predicting the level of understanding of students in English. The results of this research test get that the level of understanding of students in English is with the level of accuracy and architecture Keyword : Artificial Neural Network ; Prediction ; English; Level of Understanding; Backpropagation . English is a language that unites humans in communicating with others. The existence of language differences can make it difficult for people to understand each other in dialogue. Therefore, the role of English is very useful to unite human communication. In this case, researchers make research to predict the level of understanding of students in English. Predicting the level of understanding of students in English is needed to determine the level of ability or understanding of students in English so that students can further enhance student abilities. English is very necessary for students to support a bright future. In this study implements the Artificial Neural Network in conducting research and applying the bacpropagation method in it. To complete this study, researchers used several criteria, namely: Reading References, Hearing from the Environment, Practicing, Utilizing Technology. Of the four criteria using the backpropagation method is useful for training in predicting the level of understanding of students in English. The results of this research test get that the level of understanding of students in English is with the level of accuracy and architecture Keyword : Artificial Neural Network ; Prediction ; English; Level of Understanding; Backpropagation .

Artificial neural network implementation for masonry compressive strength estimation

An artificial neural network (ANN) implementation for the estimation of masonry compressive strength is presented. A heterogeneous sample is considered, including brick or stone elements, with cementitious or non-cementitious mortar. A multi-layer network was designed with sigmoidal neurons trained using a back-propagation algorithm. An object-oriented Java software program was developed in order to perform the training and the testing processes of the network, using real test data. The mean sum of square errors (SSE) was used as a global performance indicator of the network. The results obtained using the ANN were numerically compared with both real test data and with the results of empirical formulations. The comparisons showed that the ANN approach produced lower SSE than the considered formulations, with good performance on both heterogeneous masonry samples and different masonry systems. The presented approach could be particularly useful when little information is available, avoiding the need for invasive on-site tests and performing only laboratory tests on the brick (or stone) and the mortar. The ANN was able to predict the compressive masonry strength with a very small error, despite the heterogeneity of the considered sample.

Implementation of Cluster Analysis and Artificial Neural Networks as an Alternative for Klassen Typology and LQ: Case of Coconut

Due to the campaign supported by manufacturers of soy-oil in the United States which stated that consuming coconut oil might cause heart disease, coconuts’ popularity started to decline. As a result, the coconut was neglected by farmers and productivity was declining. Yet, recent study showed the opposite results. Coconut was good for health. As a result, the demand for coconut products increased. Indonesia, as the world’s largest coconut producers cannot maximize these opportunities due to aging coconut which have been neglected for many years. Coconut products may improve the income of farmers and encourage sustainable agriculture as well as diversify farmers’ income. Some simple methods, such as Klassen Typology or Location Quotient can be used to classify the potential area to be developed or rejuvenated. Nevertheless, these approaches are not able to generalize and not suitable to be implemented for cases with large data. This research tried to use cluster analysis such as dendrogram, Principle Component Analysis and Artificial Neural Network for classification. The results show that the dendrogram provided good results, whilst the Principle Component Analysis and Artificial Neural Networks required more data for better results.

Short-term Traffic Volume Prediction for the Merging Roads by Artificial Neural Network

Ana arterlerinde trafik sıkışıklığı yaşanan şehirlerin ikincil derecedeki yollarında trafik hacim tahminlerinin yapılması kritik konulardan biridir. Ekonomiden farklı mühendislik uygulamalarına kadar birçok alanda (satış tahminleri, endüstriyel süreç kontrolü, müşteri araştırmaları, veri doğrulama, risk yönetimi, hedef pazarlama ve inşaat mühendisliği gibi) kullanılan yapay sinir ağı (YSA) bu konuda umut verici bir çözüm olabilir. Geçmiş trafik verilerine dayanarak daha yüksek bir tahmin doğruluğu sağlayan YSA, son 30 yıldır ulaştırma mühendisliği alanındaki uygulamalarda çok popüler hale geldi. Bu makaledeki temel amaç, İstanbul'un Boğaz Köprülerinden birine katılan bir bağlantı yolunun kısa dönem trafik hacmini YSA'nın üç farklı uygulamasıyla tahmin etmektir. Bunlar İleri Besleme Geri Yayılımı (FFBP), Genelleştirilmiş Regresyon Sinir Ağı (GRNN) ve Radyal Tabanlı Fonksiyon (RBF) idi. Daha sonra elde edilen sonuçlar birbirleriyle ve Çoklu Doğrusal Regresyon (MLR) yönteminin sonuçları ile karşılaştırıldı.

Design and Implementation of ANN Based SCC with GGBS using Auromix 400

Concrete is the main building material. When concrete becomes hard, it gives strength to the structure. Many times it is a difficult task to pour the concrete into the formwork and compact it perfectly. This has been overcome by using Self-Compacting Concrete (SCC). Such a concrete is one of the advanced building materials in the field of construction industry. Unlike the other type of concrete, this kind of concrete compact’s effectively under its own weight. There is no need of any external vibration or compaction procedure to minimal the concrete in formwork. It can easily flow in every corners of the formwork without blocking. This project deals with SCC in which, the binary material used is Ground Granulated Blast Furnace Slag (GGBS) as mineral admixture at various percentage of replacement. To reduce the measure of water used in concrete, Auromix-400 is used as Super Plasticizer at a constant dosage. Several tests were carried out to study the behavior of fresh and hardened concrete. Test for fresh concrete includes slump flow, V Funnel test. Similarly, the properties of concrete were also determined by conducting compression and Spit tensile test. At the same time the simulation model was also developed to test the proposed system using the artificial neural network (ANN) protocol. The ANN model is built on six objects with multiple output-multiple. Single Output Type - In the second method, the artificial neural network model is a single input neural network that is built on top of multiple inputs - where multiple inputs has been predicted separately based on various types of neural function - Secondly, the ANN model is built on multiple inputs. The results indicate the superiority of the neural network method in terms of the accuracy of publication prediction results.

A Smart Chair Sitting Posture Recognition System Using Flex Sensors and FPGA Implemented Artificial Neural Network

Sitting is the most common status of modern human beings. Some sitting postures may bring health issues. To prevent the harm from bad sitting postures, a local sitting posture recognition system is desired with low power consumption and low computing overhead. The system should also provide good user experience with accuracy and privacy. This paper reports a novel posture recognition system on an office chair that can categorize seven different health-related sitting postures. The system uses six flex sensors, an Analog to Digital Converter (ADC) board and a Machine Learning algorithm of a two-layer Artificial Neural Network (ANN) implemented on a Spartan-6 Field Programmable Gate Array (FPGA). The system achieves 97.78% accuracy with a floating-point evaluation and 97.43% accuracy with the 9-bit fixed-point implementation. The ADC control logic and the ANN are constructed with a maximum propagation delay of 8.714 ns. The dynamic power consumption is 7.35 mW when the sampling rate is 5 Sample/second with the clock frequency of 5 MHz.

Fuzzy regression functions with a noise cluster and the impact of outliers on mainstream machine learning methods in the regression setting

The presence of outliers in the dependent and/or independent features distorts predictions with machine learning techniques and may lead to erroneous conclusions. It is important to implement methods that are robust against the outliers to make reliable predictions and to know the accuracy of the existing methods when data is contaminated with outliers. The focus of this study is to propose a robust fuzzy regression functions (FRFN) approach against the outliers and evaluate the performance of the proposed and several mainstream machine learning approaches in the presence of outliers for the regression problem. The proposed FRFN approach is based on fuzzy k-means clustering with a noise cluster. We compare the accuracy of Artificial Neural Networks (ANN), Support Vector Machines (SVM) and the proposed FRFN approaches with different training algorithms/kernel functions via simulated and real benchmark datasets. In total, accuracies of 36 ANN, SVM, and FRNF implementations with training algorithms and kernel and loss functions have been evaluated and compared to each other with samples containing outliers via a Monte Carlo simulation setting. It is observed in both Monte Carlo simulations and applications with benchmark dataset that FRFN with ANN trained with Bayes regularization algorithm and FRFN with SVM with Gaussian kernel outperforms the classical implementations of ANN and SVMs under the existence of outliers. The proposed noise cluster implementation considerably increases the robustness of fuzzy regression functions against outliers.

Sistema de Gestión de demanda eléctrica basada en la Web

The purpose of this investigation is to propose a comprehensive, innovative and technological solution about the intelligent prediction and management of electrical energy, such as the implementation of Artificial Neural Networks (ANN) used for detecting and learning patterns of behavior of nonlinear energy, including the deployment of intelligent software used to control energy management via a communication interface with the RNA. Furthermore, through the implementation of innovative computational techniques, this project aspires to develop a multi-agent endowed with artificial intelligence and knowledge of the rate for service general medium voltage with demand or 10 kW or more and modules of managents of critical electrical loads. Additionally, over time to learn new skills related to the electrical ecosystem that can be applied in an improved demand management. In this initial test phase, we have been working on the design and construction of hardware measurement devices, as well as low-cost controls and electronic instrumentation regulated by a computer system equipped with Web 2.0 architecture. The Web application proposal can analyze the value of different electrical and environmental variables, as well as schedule equipment starts and stops through communication with an electronic controller device.

Applications of Artificial Intelligence to Cryptography

This paper considers some recent advances in the field of Cryptography using Artificial Intelligence (AI) It specifically considers the applications of Machine Learning (ML) and Evolutionary Computing (EC) concepts used to generate ciphers. A short overview is given on Artificial Neural Networks (ANNs) and the principles of Deep Learning (DL) using Deep ANNs. In this context, the paper considers: (i) the implementation of EC and ANNs to generate unique and unclonable ciphers; (ii) ML strategies for detecting the genuine randomness (or otherwise) of binary streams for applications in Cryptanalysis. The paper aims to provide an overview on how AI can be applied for encrypting data and undertaking cryptanalysis of such data and other encrypted data classes in order to assess the cryptographic strength of an encryption algorithm. For example, to detect patterns of intercepted data streams that are signatures of encrypted data. An application is presented which includes authentication of high-value documents such as bank notes, using smartphones. Using an antenna of a smartphone to read (in the near field) an embedded flexible integrate circuit with a non-programmable coprocessor, ultra-strong encrypted information can be used on-line for validation.

A biohybrid synapse with neurotransmitter-mediated plasticity.

Brain-inspired computing paradigms have led to substantial advances in the automation of visual and linguistic tasks by emulating the distributed information processing of biological systems1. The similarity between artificial neural networks (ANNs) and biological systems has inspired ANN implementation in biomedical interfaces including prosthetics2 and brain-machine interfaces3. While promising, these implementations rely on software to run ANN algorithms. Ultimately, it is desirable to build hardware ANNs4,5 that can both directly interface with living tissue and adapt based on biofeedback6,7. The first essential step towards biologically integrated neuromorphic systems is to achieve synaptic conditioning based on biochemical signalling activity. Here, we directly couple an organic neuromorphic device with dopaminergic cells to constitute a biohybrid synapse with neurotransmitter-mediated synaptic plasticity. By mimicking the dopamine recycling machinery of the synaptic cleft, we demonstrate both long-term conditioning and recovery of the synaptic weight, paving the way towards combining artificial neuromorphic systems with biological neural networks.

Predictive Neural Network in Multipurpose Self-Tuning Controller

Abstract A very important problem in designing of controlling systems is to choose the right type of architecture of controller. And it is always a compromise between accuracy, difficulty in setting up, technical complexity and cost, expandability, flexibility and so on. In this paper, multipurpose adaptive controller with implementation of artificial neural network is offered as an answer to a wide range of tasks related to regulation. The effectiveness of the approach is demonstrated by the example of an adaptive thermostat. It also compares its capabilities with those of classic PID controller. The core of this approach is the use of an artificial neural network capable of predicting the behaviour of controlled object within its known range of parameters. Since such a network, being trained, is a model of a regulated system with arbitrary precision, it can be analysed to make optimal management decisions at the moment or in a number of steps. Network learning algorithm is backpropagation and its modified version is used to analyse an already trained network in order to find the optimal solution for the regulator. Software implementation, such as graphical user interface, routines related to neural network and many other, is done using Java programming language and Processing open-source integrated development environment.

SO019A PREDICTIVE MODEL OF MORTALITY IN ACUTE RENAL FAILURE IN THE CRITICAL PATIENT: USEFULNESS OF ARTIFICIAL INTELLIGENCE

Abstract Background and Aims Patients with Acute Renal Failure (ARF) have a high risk of mortality, especially those who enter the Intensive Care Unit (ICU). In this population, predictive models of mortality on prognostic scales, such as SAPS-II (Simplified Acute Physiology Score II), linearly relate risk factors without taking into account the complex relationship's variables can have. There are models where Machine Learning (ML) techniques have been used, but there is still room for improvement. The implementation of deep artificial neural networks (DANN) can be challenging. The literature models, using SAPS-II report an accuracy, f1 and ROC area (receiver operating curve) in ranges of 0.538-0.621, 0.333-0.377 and 0.781-0.809 respectively. The best results with ML are improved in neural networks of a hidden layer or random forest, being the best performance in the latter: accuracy 0.715-0.741, F1 0.449-0.470 and ROC between 0.862-0.870. The aim is to evaluate and improve the predictive capacity of ML techniques for the prediction of mortality in patients with ARF admitted to the ICU, through the use of the open database MIMIC-III (Medical Information Martfor Intensive Care III). Method Design: Retrospective analysis of historical cooperation of 20,928 patients with ARF from Beth Israel Deaconess (Boston), from 2001 to 2012. Method: ML algorithm based on DANN. Creation of a model to predict in-hospital mortality after discharge from the ICU with the variables of the first 24 hours after admission to the ICU. To evaluate the robustness of the model has been performed cross-validation by separating the samples into different combinations of training and test data (k-folds). The unavailable variables haematological with means extracted from the training set of the respective fold. The DANN trained with the complex relationship's variables folds, two hidden layers of 75 and 40 neurons respectively. Inclusion criteria: > 16 years, AKI according to KDIGO criteria and predictive variables based on SAPS-II. Variables: Age, sex, type of admission, number of admissions in ICU, heart rate, blood pressure, temperature, PaO2, FiO2, sodium, potassium, bilirubin, bicarbonate, urea, leukocytes, diuresis and diagnosis of ICD-9 (metastatic cancer, haematological malignancies). Results Accuracy: 76.7% 0 ± 1.7%, f1: 86.6% ± 1.9%, RO area: 0.859 ± 0.006, sensitivity: 75.7% ± 2.8%, specificity: 80.5% ± 2.9%. Conclusion The use of Machine Learning techniques based on deep artificial neural networks can improve the predictive ability of mortality in acute renal failure of the critically ill patient of traditional clinical risk scales and even current Machine Learning models.

Fabrication of GaOx based crossbar array memristive devices and their resistive switching properties

Memristive devices attract much attention for the application to neuromorphic computing. Here we focus on amorphous GaOx based memristive devices and their integration as a crossbar architecture which is mandatory to improve parallelism in computing and essential to future power-saving, high-speed and reliable artificial neural networks (ANN). Fabricated Pt/GaOx/ITO crossbar array memristive devices with 3 × 3 μm2 crosspoints exhibit non-filamentary reversible bipolar resistive switching (RS) without any electroforming processes. We demonstrate the gradual resistance change by repeating voltage sweep and applying pulse voltages. The expression of spike-timing-dependent plasticity property, one of synaptic functions in neuromorphic devices, is also successful in the present device. The obtained results strongly suggest that the drift of oxygen vacancies in GaOx plays an important role in the RS behavior and the GaOx based crossbar array memristive devices have a great potential for hardware implementation of ANN.

Implementation of artificial neural networks for very short range weather prediction

Weather conditions are a significant factor for various sectors such as transportation safety, development, health, etc. Therefore, high development is needed in forecasting future weather conditions. Many ways are used to predict weather conditions. Along with the development of technology now, weather prediction can be made using Artificial Intelligence (AI) technology or artificial intelligence so that the results obtained are more optimal. In this study, the artificial neural network used has a feedforward neural network algorithm using training data consisting of temperature, air pressure, air humidity, wind speed, hourly wind speed at the Juanda Meteorological Station in Surabaya in Januari 2019 with the target is intensity of rainfall. Furthermore, the data was released in the period of 1 January 2019 to 31 Januari 2019. Based on the analysis results, the Artificial Neural Network model has a fairly good performance in predicting an increase in rainfall in Surabaya. The best model is considered by a model with architecture 7 - 60 - 1 with an estimate correlation is 0.87, with an error value of -0.03. With this model, it is expected to become one of the forecaster considerations in making special weather forecasts at intervals every hour.

Implementation of backpropagation artificial neural network for early detection of vitamin and mineral deficiency

Basic Health Research data in 2018 show that around 95.5 percent of Indonesian people less consumption of fruit and vegetables. This condition then leads to the suspicion of vitamin and mineral deficiency in Indonesia people. Several methods have been used to detect vitamin and mineral deficiency, such as convolutional rule-based and certainty factor method. However, these methods are less adaptive to adapt to the changes in symptoms when detecting vitamin and mineral deficiencies. This paper proposes an artificial neural network (ANN) using backpropagation (BPN) to detect the vitamin and mineral deficiencies in the human body. Using 107 input of physical symptoms and 17 output of the type of vitamin and mineral, the architecture of the ANN consist of 107-50-17 neurons for the input layer, hidden layer, and output layer respectively. Based on some trial and error experiments, can be determined the epoch, the learning rate, and the error rate to produce the optimal result of the detection. This experiment using 623 epochs, 0.0517 error rate, and 0.1 for the learning rate. The performance measurement conducted using precision, recall, and F-score, for each class output. The experiment shows the proposed ANN using BPN reaches an accuracy level of 73%.

Pattern Recognition Techniques for the Identification of Activities of Daily Living Using a Mobile Device Accelerometer

The application of pattern recognition techniques to data collected from accelerometers available in off-the-shelf devices, such as smartphones, allows for the automatic recognition of activities of daily living (ADLs). This data can be used later to create systems that monitor the behaviors of their users. The main contribution of this paper is to use artificial neural networks (ANN) for the recognition of ADLs with the data acquired from the sensors available in mobile devices. Firstly, before ANN training, the mobile device is used for data collection. After training, mobile devices are used to apply an ANN previously trained for the ADLs’ identification on a less restrictive computational platform. The motivation is to verify whether the overfitting problem can be solved using only the accelerometer data, which also requires less computational resources and reduces the energy expenditure of the mobile device when compared with the use of multiple sensors. This paper presents a method based on ANN for the recognition of a defined set of ADLs. It provides a comparative study of different implementations of ANN to choose the most appropriate method for ADLs identification. The results show the accuracy of 85.89% using deep neural networks (DNN).

Artificial Neural Network Implementation for Classification of Lung Tissues in High Resolution Tomography Images

In this digital world, the amount of information what we are getting is uncountable and speed at which it reaches us is unpredictable. We are getting is uncountable. In order to understand and predict the required data we should have the image classification technique otherwise there will be lot of confusion and misunderstanding will happen during its study. Content based image retrieval technique can be used to exploit the properties of high resolution computed tomography data which will be stored in the achieve by identifying the similar images which will helped for radiologist for self learning and differential diagnosis of intestinal lungs diseases. This technique identifies the particular image based on the content present in the image rather than keywords or tag related to the image. The content present in the image means, patterns, colors, textures etc. high resolution technique of image identification method is used in many medical applications like identifying diseases or the identification of symptoms of the diseases. in this project we are using the neural network classifier to classify the lung tissues in high resolution computed tomography images. This method gives the proper and clear classification of the lung tissues which will helps in medical science field. This neural network classifier uses the optimal subset feature to classify the pattern of the lung tissues.

Optically Modulated Threshold Switching in Core–Shell Quantum Dot Based Memristive Device

AbstractThe threshold switching (TS) phenomenon in memristors has drawn great attention for its versatile applications in selectors, artificial neurons, true random number generators, and electronic integrations. The transition between nonvolatile resistive switching and volatile TS modes can be realized by doping, varying annealing and voltage sweeping conditions, or imposing different compliance current. Here, a strategy is reported to achieve such transition by the noninvasive UV light stimulus based on InP/ZnS quantum dot (QD) memristor. The core–shell InP/ZnS QDs with quasi‐type II band alignment ensures photoexcited electrons localized in InP core, photoexcited hole state distributed in the outer shell, and subsequent lifetime controlling of conductive filament under light irradiation. Systematic mechanism investigations indicate that UV photogenerated holes are accumulated on the surface of the QD film, which is consistent with rapid transfer of photogenerated holes in the core–shell InP/ZnS structure. Based on the light‐modulated effect, a reconfigurable 9 × 9 visual data storage array with a key pattern and a simple leaky integrate‐and‐fire circuit are constructed. These results suggest the potential of direct optical modulation of memory mode through energy band engineering, leading to future optoelectronic and electronic device for the implementation of neuromorphic visual system and artificial neural networks.

Cholesky–ANN models for predicting multivariate realized volatility

AbstractAccurately forecasting multivariate volatility plays a crucial role for the financial industry. The Cholesky–artificial neural networks specification here presented provides a twofold advantage for this topic. On the one hand, the use of the Cholesky decomposition ensures positive definite forecasts. On the other hand, the implementation of artificial neural networks allows us to specify nonlinear relations without any particular distributional assumption. Out‐of‐sample comparisons reveal that artificial neural networks are not able to strongly outperform the competing models. However, long‐memory detecting networks, like nonlinear autoregressive model process with exogenous input and long short‐term memory, show improved forecast accuracy with respect to existing econometric models.