Keras Library Research Articles

Development of the trainer program based on neural network technologies for development of spatial thinking of students

Introduction. For the last thirty years, UNESCO has been repeatedly raising questions of accessibility and quality of education for every child, teen, and adult by implementing the “Education for All” program. The new concept of education proposed by UNESCO aims at transforming people’s lives and formulates the following sustainable development goal “To provide inclusive and equitable quality education and create lifelong learning opportunities for all. One strategy for achieving this goal is to use and develop information and communication technology (ICT) to strengthen education systems, knowledge dissemination, access to information, and quality and effective learning. The aim of the work is to increase the efficiency of students’ learning of geometrical and graphic disciplines by developing a pilot electronic course in LMS Moodle and special software implementing elements of individualization of the learning process. Materials and methods. Theoretical research methods included generalization of scientific and pedagogical literature on formation of spatial thinking. Empirical research methods included testing, interviewing, and questioning using Google forms. Mathematical methods of research included algorithmizing and programming. Python 3.9 language and PyCharm development environment were used for writing the trainer program. The OpenCV, Matplotlib and Keras libraries, Microsoft PowerPoint, Microsoft Visio and CAD Compass-3D were used to create methodological materials Results. The software simulator for developing spatial thinking of engineering students studying “Descriptive Geometry” discipline has been developed in three versions. Conclusion. The ideas about the level of spatial thinking of modern students are supplemented. The original programs for the implementation of the elements of individualization of the educational process in the discipline of “Descriptive geometry” and algorithms of its work were developed and introduced into the scientific turnover, as well as the proposed pedagogical conditions of their use. Practical significance lies in the creation of learning and teaching methods and program basic elements for the use of end-to-end digital technology in the learning process.

Automated Protein Secondary Structure Assignment from Cα Positions Using Neural Networks.

The assignment of secondary structure elements in protein conformations is necessary to interpret a protein model that has been established by computational methods. The process essentially involves labeling the amino acid residues with H (Helix), E (Strand), or C (Coil, also known as Loop). When particular atoms are absent from an input protein structure, the procedure becomes more complicated, especially when only the alpha carbon locations are known. Various techniques have been tested and applied to this problem during the last forty years. The application of machine learning techniques is the most recent trend. This contribution presents the HECA classifier, which uses neural networks to assign protein secondary structure types. The technique exclusively employs C coordinates. The Keras (TensorFlow) library was used to implement and train the neural network model. The BioShell toolkit was used to calculate the neural network input features from raw coordinates. The study’s findings show that neural network-based methods may be successfully used to take on structure assignment challenges when only C trace is available. Thanks to the careful selection of input features, our approach’s accuracy (above 97%) exceeded that of the existing methods.

Devanagari Character Recognition Using Deep Learning

The Devanagari script is one of the broadly used scripts of India and is advanced from the Brahmi script. It includes forty-seven number one alphabets, 14 vowels, 33 consonants, and 10 digits. There isn't any capitalization of letters, like in the English language. The Devanagari script includes consonants and modifiers. Deep gaining knowledge of strategies is carried out to extract capabilities and apprehend the characters in an image. A Deep Convolutional Neural Network (DCNN) had been included to extract capabilities and classify the entered images. Consecutive convolutional layers are used in this manner which brings introduced gain withinside the manner of extracting higher-degree capabilities. The output of the CNN layers is fed to the completely related layers. The version has applied the use of Keras libraries on the pinnacle of a TensorFlow backend. Finally, the threat or possibility rating of every person is decided and the person with the very best possibility rating is proven because of the output. Handwritten Devanagari Character Recognition is extra tough in contrast to the popularity of the Roman characters.

Classification of nitrogen deficiency for maize plants using deep learning algorithms on low-end android smartphones

Maize comprises about one-fifth of the calories eaten in Sub-Saharan Africa. However, farmers in Sub-Saharan Africa have been unable to produce enough food for consumption. This is mostly due to a shortage of nutrients and the adoption of antiquated agronomic methods. Nitrogen deficiency or deficiency is a significant factor contributing to this poor yield of maize crop. Many smallholder farmers lack the necessary information to recognize this nitrogen deficit early on, when it is still reversible, before it damages their fields and maize yields. The purpose of this project is to determine a way for developing a mobile app for low-end Android phones that use a machine learning model to detect nitrogen insufficiency. The model is constructed by utilizing the Tensorflow and Keras libraries to train a pre-trained Single Shot Detector (SSD) Mobilenet model. Additionally, the approach takes advantage of Keras's built-in Image Augmentation algorithms to produce additional photos for our datasets. The model generated is 81 percent accurate. The Android application enables a smallholder farmer to maintain and analyze the soil health of several farms, as well as to determine the necessary fertilizer application to help rectify the nitrogen shortfall. Future directions in this field of study have also been highlighted for the benefit of interested scholars.

CARD9: Cardiac Contractility Estimation Based on Left Ventricular Pressure: Comparison of Time Series Classifiers and Graph-based Neural Networks

Background: In heart failure therapy, the real-time estimation of remaining cardiac contractility (RCC) can facilitate patient management, as well as, the performance of physiological controllers for ventricular assist devices (VAD). In this study, the real-time estimation of RCC based on time series data (TSD) of the left ventricular pressure (LVP) was investigated by exploiting two traditional time series classifiers (TSC) and two graph-based neural networks (GNN). All classifiers were assessed with respect to their RCC estimation accuracy and their applicability for real-time VAD control. Methods: For the RCC estimation (Figure 1), LVP TSD was generated using a hybrid mock circulation (HMC). On the HMC instantaneous pressure, volume, and flowrate values were computed by a numerical model of the human cardiovascular system and applied by a hydraulic interface on the inlet and outlet mixed-flow blood pump. The experimental protocol for the acquisition of the LVP TSD included 63 experiments, where preload, afterload, pump speed, and heart rate variations were performed for 9 RCC values. The data was pre-processed by segmenting and normalizing the TSD, producing 6300 cardiac cycle samples, which were used for the TSC approaches. The TSCs comprised the dynamic time warping nearest neighbor (DTW-NN) classifier and the support vector (SVM) classifier. Furthermore, the maturity of two GNN approaches was explored, including a simple custom architecture (C-GNN) and a pretrained architecture (P-GNN) provided by the KERAS library (Inception V3). The C-GNN comprised a convolutional layer with 32 filters of size 3x3, a max pooling layer with filter size 2x2, a dense layer with 128 units, and a dense layer with 9 units for the classification output. For the TSD to be used in the GNN frameworks, two different image encoders (IE), namely direct plot and recurrence plot encoders, were used and assessed. The data was split 80/20 for network training and validation. Results: All four classifiers, DTW-NN, SVM, C-GNN (IE: recurrence plot), and the P-GNN (IE: direct plot) achieved an accuracy of at least 98%. The SVM had the highest accuracy (99.9%) and the shortest prediction time (63 μs/sample). The prediction time of DTW-NN was on average 4.84 s/sample, prohibiting the implementation in real-time VAD control. Both GNN architectures achieve high accuracy and short prediction time, however, their performance does not reach the benchmark set by the SVM. A comparison of the image encoders showed that both the direct plot and the recurrence plot encoder lead to similar classification performances when using the same GNN, but different performances between the two investigated GNNs. Conclusion: All classification methods provided accurate RCC estimation with the SVM showing superior performance and being the most promising for real-time clinical implementation. These estimation approaches could substantially support patient surveillance and physiologic VAD control approaches. Figure 1. Schematic of the sequential steps for the estimation of the remaining cardiac contractility from time-series data (TSD) of the left ventricular pressure (LVP). a) LVP TSD is produced through a hybrid mock circulation, where the human cardiovascular system is simulated and the pressures, volume, and flowrates calculated are applied on the inlet and outlet of a ventricular assist device in real-time via a hydraulic interface. LVP TSD was acquired for various physiologic conditions. b) Pre-processing of the data, including cardiac cycle segmentation and time normalization. c) Translation of the TSD to images through different image encoders. d) Time series classifiers, including dynamic time warping nearest neighbor (DTW-NN) and the support vector (SVM) classifiers. e) Graph-based neural networks (GNN). f) Classified cardiac contractility.

A Gentle Introduction to Physics-Informed Neural Networks, with Applications in Static Rod and Beam Problems

A modern approach to solving mathematical models involving differential equations, the so-called Physics-Informed Neural Network (PINN), is based on the techniques which include the use of artificial neural networks and the method of fitting the governing differential equations at collocation points. In this paper, training of the PINN with an application of optimization techniques is performed on simple one-dimensional mechanical problems of elasticity, namely rods and beams. Different boundary conditions are considered. Required computer algorithms are implemented using Python programming packages with the intention of creating neural networks. Numerical results are presented, and the efficiency of the proposed technique is investigated through numerical experiments with different numbers of epochs, batches, hidden layers, neurons, and collocation points. The paper provides useful skills for using a PINN for different problems of solid mechanics. The proposed methodology is a continuation of our intention of using PINNs for problems of the theory of elasticity. The objectives are to present simply the main steps of constructing PINN and an implementation of it. A detailed explanation of the Python programming code, based on the scientific software Tensorflow, built in the Keras library and optimizers, may help compose an effective code for complicated models in mechanics. PINNs are proposed in many recent publications to solve complicated direct and inverse problems. It seems to be a promising method that will play a central role in the development of computational mechanics in the near future. Nevertheless, the lack of educational material does not help new users to enter this scientific area. The present contribution describes the method for the solution of elementary rod and beam problems and gives computer codes that may help the reader to understand the method and to apply it to other problems.

Comparison of Models for Wind Speed Prediction Through Neural Networks in Lençóis, Bahia

This study aims to compare forecasting models using artificial intelligence to conclude which is the best one for the forecasting wind speed for 1 hour in Lençóis, BA, using a data source from the Instituto Nacional de Meteorologia (INMET). Furthermore, an Artificial Neural Network (ANN) was developed using TensorFlow and Keras libraries, it was compared with other forecasting models, which showed to be the most efficient among the options for this purpose. Moreover, the principal metric used to evaluate this study was Mean Absolute Error (MAE), and the auxiliary ones were Mean Squared Error (MSE) and R². The RNA obtained the following values for each metric: 0.421 for MAE, 0.389 for MSE, and 0.523 for the R² metric.

Pendeteksi Sampah Metal untuk Daur Ulang Menggunakan Metode Convolutional Neural Network

Waste is part material that has no value within the scope of production. If you no longer need it, metal cans can take about 80 to 200 years to decompose. CNN is part of the supervised learning method that exists in deep learning, where those who have expertise in representing images or images from several categories increase recognition, namely in classifying objects, doing scene recognition, and detecting object detection. In this study, using the CNN method as a development model and applying the ResNet 50 network design, which includes the type Convolutional Neural Network (CNN) that operates by way of working, namely receive an input in the form of an image or images. The input will be carried out by training that is set using the CNN architecture so that later it will produce an output that can recognize objects as expected in knowing the types of cardboard and glass waste. The implementation of this research uses the Python programming language, Anvil, and the TensorFlow and Keras libraries. The system has succeeded in detecting the type of metal waste from general waste and assisting third parties, namely implementing it through the website using Anvil. The input shape for CNN modeling in this study is 512x384 pixels, which has a value of 100 eras, and the data set used contains images of metal waste and general waste found 547 images, resulting in an accuracy of 96%.

Development of Adaptive Testing Method Based on Neurotechnologies

Purpose of the study. The aim of the study is to create neural network models of modules in an adaptive testing system to design an individual testing trajectory.The research article discusses the implementation of an adaptive testing system in terms of introducing artificial neural network modules into its composition, which should solve the problem of choosing a topic and the complexity of the next question, taking into account previous answers and the complexity of previously asked questions, as well as the connectivity of topics and response time as a factor guessing or searching for an answer, thereby forming an individual testing trajectory.Materials and methods. In the course of the study, the data that affect the quality of the solution of the problem was analyzed, the general modular structure of the system was proposed, and the main data flows entering the input of an artificial neural network (ANN) were described. To solve the problem of choosing the complexity of a question, it is proposed to use a feed-forward network, a comparison of various ANN architectures and training parameters (weight update algorithms, loss functions, number of training epochs, packet sizes) is carried out. As an alternative, the possibility of using a recurrent ANN LSTM (Long-Short Term Memory) network is considered. All results were obtained using the high-level Keras library, which allows you to quickly start at the initial stages of research and get the first results. SGD, Adam, NAdam and RMSprop implemented in Keras were compared as optimizers to achieve faster convergence. Adam showed the best results in terms of accuracy, while the MSE loss function (mean square error) was used together with the optimizer. Traditionally, training was carried out for a large number of epochs; graphs of dependences of accuracy on the number of epochs for a different number of neurons in the hidden layer were experimentally obtained.Results. Based on the study, we can conclude that the obtained accuracy of the direct propagation network of 80-85% is quite sufficient for its use in the adaptive testing system. However, it remains to answer the question of the need to improve the efficiency of an already implemented network, and, therefore, to conduct research on methods to improve the efficiency of networks, including finer tuning of parameters and learning algorithms, as well as architecture.A well-known and obvious drawback of using LSTMs is their exactingness in terms of equipment and resources, both during training (the training process takes a significant amount of time) and during startup, in our case, it is supplemented by increased requirements for the training sample and casts doubt on the advisability of further study of LSTM networks when solving this task.Conclusion. The introduction of the proposed tools will allow implementing an adaptive testing system, with an intelligent selection of questions depending on the demonstrated level of knowledge of the test person to form an individual testing trajectory in order to determine the reliable level of knowledge of the test subject for the optimal number of questions asked.

Bitcoin Price Prediction Using Machine Learning Techniques

This paper discusses, trying to accurately assess the price of Bitcoin by looking at different parameters affects the value of Bitcoin. In our work, we focus on understanding and seeing the evolution of Bitcoin daily market, a1 and gaining intuition in the most relevant aspects surrounding the Bitcoin price. In the meantime, market capitalization of publicly traded cryptocurrencies exceeds $ 230 billion. The most important cryptocurrency, Bitcoin, is used primarily as a digital value store, and its pricing opportunities have been extensively considered. These features are described in more detail in the following paragraph: details of the main Bitcoin, as described in the paper. Bitcoin is the most expensive digital currency in the market. However, Bitcoin prices have been highly volatile, making it difficult to forecast. As a result, the goal of this research is to find the most efficient and accurate model for predicting Bitcoin prices using various machine learning algorithms. Several regression models with scikit-learn and Keras libraries were tested using 1-minute interval trading data from the Bitcoin exchange website bit stamp from January 1. 2012 to January 8, 2018. The best results showed a Mean Squared Error (MSE) as low as 0.00002 and an R- Square (R2) as high as 99.2 percent.

Generative time series model based on encoder-decoder architecture

Encoder-decoder neural network models have found widespread use in recent years for solving various machine learning problems. In this paper, we investigate the variety of such models, including the sparse, denoising and variational autoencoders. To predict non-stationary time series, a generative model is presented and tested, which is based on a variational autoencoder, GRU recurrent networks, and uses elements of neural ordinary differential equations. Based on the constructed model, the system is implemented in the Python3 environment, the TensorFlow2 framework and the Keras library. The developed system can be used for modeling continuous time-dependent processes. The system minimizes a human factor in the process of time series analysis, and presents a high-level modern interface for fast and convenient construction and training of deep models.

Tuning of data augmentation hyperparameters in deep learning to building construction image classification with small datasets.

Deep Learning methods have important applications in the building construction image classification field. One challenge of this application is Convolutional Neural Networks adoption in a small datasets. This paper proposes a rigorous methodology for tuning of Data Augmentation hyperparameters in Deep Learning to building construction image classification, especially to vegetation recognition in facades and roofs structure analysis. In order to do that, Logistic Regression models were used to analyze the performance of Convolutional Neural Networks trained from 128 combinations of transformations in the images. Experiments were carried out with three architectures of Deep Learning from the literature using the Keras library. The results show that the recommended configuration (Height Shift Range = 0.2; Width Shift Range = 0.2; Zoom Range =0.2) reached an accuracy of 95.6% in the test step of first case study. In addition, the hyperparameters recommended by proposed method also achieved the best test results for second case study: 93.3%.

An Intelligent AI for Transportation System

Public transportation is used by many people all over the world to visit new places. Individuals are in general make a plan to trip new places and for special occasions they might feel totally lost in new place. At this time our Chatbot acts the helper for travellers. Our Chatbot helps the people to find their desire source and destination. It is a application that is work on web, desktop and also on telegram application. This Chatbot uses the Natural Language Processing(NLP) and Deep Learning methods. It acts as the live human being as a Graphical User Interface(GUI). This AI Chatbot ask user for current location and destination location by asking some questions. It analyze the user’s queries and associate with data set of the Chatbot. The deep learning techniques are used in this Chatbot are responsible for understanding the user’s questions perfectly to avoid any misunderstanding. Once the users question has been understand, then the Chatbot provides the most appropriate reply for the users query request. Then the users gets all the information related to the bus names with their numbers so the person can travel to the desired location. Our Chatbot implemented by using Pythons Keras library and Tkinter is used for GUI purpose.

Accuracy of advanced deep learning with tensorflow and keras for classifying teeth developmental stages in digital panoramic imaging

BackgroundThis study aims to propose the combinations of image processing and machine learning model to segment the maturity development of the mandibular premolars using a Keras-based deep learning convolutional neural networks (DCNN) model.MethodsA dataset consisting of 240 images (20 images per stage per sex) of retrospect digital dental panoramic imaging of patients between 5 and 14 years of age was retrieved. In image preprocessing, abounding box with a dimension of 250 × 250 pixels was assigned to the left mandibular first (P1) and second (P2) permanent premolars. The implementation of dynamic programming of active contour (DP-AC) and convolutions neural network on images that require the procedure of image filtration using Python TensorFlow and Keras libraries were performed in image segmentation and classification, respectively.ResultsImage segmentation using the DP-AC algorithm enhanced the visibility of the image features in the region of interest while suppressing the image's background noise. The proposed model has an accuracy of 97.74%, 96.63% and 78.13% on the training, validation, and testing set, respectively. In addition, moderate agreement (Kappa value = 0.58) between human observer and computer were identified. Nonetheless, a robust DCNN model was achieved as there is no sign of the model's over-or under-fitting upon the learning process.ConclusionsThe application of digital imaging and deep learning techniques used by the DP-AC and convolutions neural network algorithms to segment and identify premolars provides promising results for semi-automated forensic dental staging in the future.

Review of Deep Transfer Learning Models for Image Classification

With the rapid rise in urbanization, solid waste generation has increased exceedingly. This study aims to develop a Convolutional Neural Network (CNN) to classify trash into biodegradable and non-biodegradable wastes. The TrashNet dataset was utilized in this study, and image augmentation was employed to make the model more robust against translation invariance. Transfer Learning methods based on CNN have shown promising outcomes on diverse image classification problems. This paper reviews the deep learning models available with pre-trained weights in the Keras library. The performance of the models was compared, and the model based on NASNetMobile had the highest accuracy of 97%. Further, the model’s hyper-parameters were tuned, and the significance of a hyper-parameter on the model’s accuracy was studied.

Body Height Estimation In Irrigation Dams With Deep Learning Model

Dams are one of the most important constructions for our country. The body height of the dams is one of the important factors in the efficiency of the dams. Today, the body height of dams is calculated by engineers. The aim of our study is to calculate the dam height with the deep learning model of artificial intelligence. Modeling was coded with python software. Numpy pandas libraries were used for the analysis of dam data. Matplotlib and seaborn were employed to visualize the data. Sklearn, tensorflow and keras libraries were used for deep learning modeling. Dam data are limited to irrigation dams in Turkey. For data analysis, the altitude, height, volume, area, temperature and precipitation characteristics were taken into consideration. As a result of our study, the dam body height estimation was done by teaching the dam data to the machine through multi-layer artificial neural networks of the deep learning model. The deviation in the body height estimations was found to be higher due to the insufficient data.

Assistive Technology-Based Solution for Hearing Impairment Using Smartphones

Developing a system for sign language recognition becomes essential for the deaf as well as a mute person. The recognition system acts as a translator between a disabled and an able person. This eliminates the hindrances in the exchange of ideas. Most of the existing systems are very poorly designed with limited support for the needs of their day to day facilities. The proposed system embedded with gesture recognition capability has been introduced here which extracts signs from a video sequence and displays them on screen. On the other hand, a speech to text as well as text to speech system is also introduced to further facilitate the grieved people. To get the best out of a human-computer relationship, the proposed solution consists of various cutting-edge technologies and Machine Learning based sign recognition models that have been trained by using TensorFlow and Keras library. The proposed architecture works better than several gesture recognition techniques like background elimination and conversion to HSV

BITCOIN PRICE PREDICTION USING SVM

Bitcoin is one of the most valuable Crypto currency in the world with the prices as high as 68,078 United States Dollar (USD) in November of 2021. It made Bitcoin a very profitable market for investment but Bitcoin saw many ups and down as well. Prices of Bitcoin have highly fluctuated which make them very difficult to predict. Hence, this research aims to discover the most efficient and highest accuracy model to predict Bitcoin prices from various machine learning algorithms. By using 1-minute interval trading data on the Bitcoin exchange website named bitstamp some different regression models with Scikit- learn and Keras libraries had experimented. The dataset used contains minute by minute prices of Bitcoin of over 5 years and contains almost 3 Million entries. Since, the dataset used is a big data, evaluating the performance of algorithms over a large dataset will give accurate results.

Impact of the Radar Image Resolution of Military Objects on the Accuracy of their Classification by a Deep Convolutional Neural Network

Introduction. Deep convolutional neural networks are considered as one of the most promising tools for classifying small-sized objects on radar images. However, no systemic study has been reported so far on the dependence between the classification accuracy achieved by convolutional neural networks and such an important image characteristic as resolution.Aim. Determination of a dependence between of the accuracy of classifying military objects by a deep convolutional neural network and the resolution of their radar images.Materials and methods. An eight-layer convolutional neural network was designed, trained and tested using the Keras library and Tensorflow 2.0 framework. For training and testing, the open part of the standard MSTAR dataset comprising ten classes of military objects radar images was used. The initial weight values of the MobileNetV1 and Xception networks used for a comparative assessment of the achieved classification accuracy were obtained from the training results on the Imagenet.Results. The accuracy of classifying military objects decreases rapidly along with a deterioration in resolution, amounting to 97.91, 90.22, 79.13, 52.2 and 23.68 % at a resolution of 0.3, 0.6, 0.9, 1.5 and 3 m, respectively. It is shown that the use of pretrained MobileNetV1 and Xception networks does not lead to an improvement in the classification accuracy compared to a simple VGG-type network.Conclusion. Effective recognition of military objects at a resolution worse than one meter is practically impossible. The classification accuracy of deep neural networks depends significantly on the difference in the image resolution of the training and test sets. A significant increase in the resistance of the classification accuracy to changes in the resolution can be achieved by training on a set of images with different resolutions.

Detecting Damaged Buildings on Post-Hurricane Satellite Imagery based on Transfer Learning

In this article, Convolution Neural Network (CNN) is used to detect damage and no damage images form satellite imagery using different classifiers. These classifiers are well-known models that are used with CNN to detect and classify images using a specific dataset. The dataset used belongs to the Huston hurricane that caused several damages in the nearby areas. In addition, a transfer learning property is used to store the knowledge (weights) and reuse it in the next task. Moreover, each applied classifier is used to detect the images from the dataset after it is split into training, testing and validation. Keras library is used to apply the CNN algorithm with each selected classifier to detect the images. Furthermore, the performance measures are used as a criterion to decide which classifier is the best one to detect the images with high accuracy. Eventually, the simulation results show that each classifier detect the damage/no damage image with different performance measures and then makes it easy to select the best one.