Narrow Results

Black gram crop belongs to the Fabaceae family and its scientific name is Vigna Mungo.It has high nutritional content, improves the fertility of the soil, and provides atmospheric nitrogen fixation in the soil. The quality of the black gram crop is degraded by diseases such as Yellow mosaic, Anthracnose, Powdery Mildew, and Leaf Crinkle which causes economic loss to farmers and degraded production. The agriculture sector needs to classify plant nutrient deficiencies in order to increase crop quality and yield. In order to handle a variety of difficult challenges, computer vision and deep learning technologies play a crucial role in the agricultural and biological sectors. The typical diagnostic procedure involves a pathologist visiting the site and inspecting each plant. However, manually crop disease assessment is limited due to lesser accuracy and limited access of personnel. To address these problems, it is necessary to develop automated methods that can quickly identify and classify a wide range of plant diseases. In this paper, black gram disease classifications are done through a deep ensemble model with optimal training and the procedure of this technique is as follows: Initially, the input dataset is processed to increase its size via data augmentation. Here, the processes like shifting, rotation, and shearing take place. Then, the model starts with the noise removal of images using median filtering. Subsequent to the preprocessing, segmentation takes place via the proposed deep joint segmentation model to determine the ROI and non-ROI regions. The next process is the extraction of the feature set that includes the features like improved multi-texton-based features, shape-based features, color-based features, and local Gabor X-OR pattern features. The model combines the classifiers like Deep Belief Networks, Recurrent Neural Networks, and Convolutional Neural Networks. For tuning the optimal weights of the model, a new algorithm termed swarm intelligence-based Self-Improved Dwarf Mongoose Optimization algorithm (SIDMO) is introduced. Over the past two decades, nature-based metaheuristic algorithms have gained more popularity because of their ability to solve various global optimization problems with optimal solutions. This training model ensures the enhancement of classification accuracy. The accuracy of the SIDMO, which is around 94.82%, is substantially higher than that of the existing models, which are FPA$=$88.86%, SSOA$=$88.99%, GOA$=$85.84%, SMA$=$85.11%, SRSR$=$85.32%, and DMOA$=$88.99%, respectively.

In the sign language fingerspelling scheme, letters in the alphabet are presented by a distinctive finger shape or movement. The presented work is conducted for autokinetic translating fingerspelling signs to text. A recognition framework by using intensity and depth information is proposed and compared with some distinguished works. Histogram of Oriented Gradients (HOG) and Zernike moments are used as discriminative features due to their simplicity and good performance. A Deep Belief Network (DBN) composed of three Restricted Boltzmann Machines (RBMs) is used as a classifier. Experiments are executed on a challenging database, which consists of 120,000 pictures representing 24 alphabet letters over five different users. The proposed approach obtained higher average accuracy, outperforming all other methods. This indicates the effectiveness and the abilities of the proposed framework.

In recent years, a robust increasing interest has been observed in wearable devices featuring smart health, smart fitness, and human–machine interaction applications. While we gained some advances on use of surface electromyography (sEMG) signals recorded from upper extremities for controlling external devices, only limited attempt has been made to track the status of targeted muscles and forecast muscle fatigue onset. In this study, we address use of sEMG signals acquired from upper extremities to predict onset of muscle fatigue using deep belief networks (DBNs) as a learning mechanism. We demonstrate that a deep architecture can learn from raw data and provide comparable performance to feature-based approaches. Experimental results show that the DBNs model investigated in this study achieves an average classification accuracy of 85.3% without any subject-oriented calibration and achieves a best case accuracy of 97.60%. A transient-to-fatigue state is introduced before the fatigue onsets as an early warning state. The aim of this paper is to evaluate the performance of the popular deep models in real fatigue detection applications. The model provides a promising result compared with state-of-art works without any feature selection process, which could potentially generate better features while reducing the requirement for expertise in data.

In this paper, a novel approach based on deep belief networks (DBN) for electrocardiograph (ECG) arrhythmias classification is proposed. The construction process of ECG classification model consists of two steps: features learning for ECG signals and supervised fine-tuning. In order to deeply extract features from continuous ECG signals, two types of restricted Boltzmann machine (RBM) including Gaussian–Bernoulli and Bernoulli–Bernoulli are stacked to form DBN. The parameters of RBM can be learned by two training algorithms such as contrastive divergence and persistent contrastive divergence. A suitable feature representation from the raw ECG data can therefore be extracted in an unsupervised way. In order to enhance the performance of DBN, a fine-tuning process is carried out, which uses backpropagation by adding a softmax regression layer on the top of the resulting hidden representation layer to perform multiclass classification. The method is then validated by experiments on the well-known MIT-BIH arrhythmia database. Considering the real clinical application, the inter-patient heartbeat dataset is divided into two sets and grouped into four classes (N, S, V, F) following the recommendations of AAMI. The experiment results show our approach achieves better performance with less feature learning time than traditional hand-designed methods on the classification of ECG arrhythmias.

Near infrared spectroscopy (NIRS) analysis technology, combined with chemometrics, can be effectively used in quick and nondestructive analysis of quality and category. In this paper, an effective drug identification method by using deep belief network (DBN) with dropout mechanism (dropout-DBN) to model NIRS is introduced, in which dropout is employed to overcome the overfitting problem coming from the small sample. This paper tests proposed method under datasets of different sizes with the example of near infrared diffuse reflectance spectroscopy of erythromycin ethylsuccinate drugs and other drugs, aluminum and nonaluminum packaged. Meanwhile, it gives experiments to compare the proposed method’s performance with back propagation (BP) neural network, support vector machines (SVMs) and sparse denoising auto-encoder (SDAE). The results show that for both binary classification and multi-classification, dropout mechanism can improve the classification accuracy, and dropout-DBN can achieve best classification accuracy in almost all cases. SDAE is similar to dropout-DBN in the aspects of classification accuracy and algorithm stability, which are higher than that of BP neural network and SVM methods. In terms of training time, dropout-DBN model is superior to SDAE model, but inferior to BP neural network and SVM methods. Therefore, dropout-DBN can be used as a modeling tool with effective binary and multi-class classification performance on a spectrum sample set of small size.

Virtual screening (VS) is defined as the use of a compilation of computational procedures to grade, score and/or sort several chemical formations. The purpose of VS is to identify the molecules holding the greatest prior probabilities of activity. Many of the conventional similarity methods assume that molecular features that do not relate to the biological activity carry the same weight as the important ones. For this reason, the researchers on this paper investigated that some features are being more important than others through the chemist structure diagrams and the weight for each fragment should be taken into consideration by giving more weight to those fragments that are more important. In this paper, a deep learning method specifically known as Deep Belief Networks (DBN) has been used to reweight the molecule features and based on this new weigh, the reconstruction feature error has been calculated for all the features. Based on the reconstruction feature error values, Principal Component Analysis (PCA) has been used for the dimension’s reduction and only few hundreds of features have been selected based on the less error rate. The main aim of this research is to show an improvement of the similarity searching performance based on the selected features those have less error rate. The results derived through the DBN were compared with those derived through other similarity methods, such as the Tanimoto coefficient and the quantum-based methods. This comparison revealed the performance of the DBN with the structurally heterogeneous data sets (DS1 and DS3) to be superior to the performances of all the other techniques.