Because of retina abnormalities of diabetic patients, the most common vision-threatening disease is diabetic retinopathy (DR). The DR diagnosis and prevention are challenging tasks as they may lead to vision loss. Acc...
详细信息
Because of retina abnormalities of diabetic patients, the most common vision-threatening disease is diabetic retinopathy (DR). The DR diagnosis and prevention are challenging tasks as they may lead to vision loss. According to the literature analysis, the shortcomings in existing studies, such as failed to reduce the feature dimension, higher execution time, and higher computational cost, unable to tune the hyper-parameters, such as a number of hidden layers and learning rate, more computational complexities, higher cost, and so forth, during DR classification. To tackle these problems, we proposed a deep long- and short-term memory (LSTM) in a neural network with redfoxoptimization (deep LSTM-RFO) algorithm for DR classification. The four major components involved in the proposed methods are image preprocessing, segmentation, feature extraction, and classification. At first, an adaptive histogram equalization and histogram equalization model performs the fundus image preprocessing, thereby neglecting the noise and improving the contrast level of an image. Next, an adaptive watershed segmentation model effectively segments the lesion region based on the optic disc color and size of hemorrhages. At the third stage, we have extracted statistical, intensity, color, and shape features. Finally, the single normal class with three abnormal classes such as mild non-proliferative diabetic retinopathy, moderate NPDR, and severe NPDR are accurately classified using the deep LSTM-RFO algorithm. Experimentally, the MESSIDOR, STARE, and DRIVE datasets are used for both training and validation. MATLAB software performs the implementation process with respect to various evaluation criteria used. However, the proposed method accomplished superior performance, such as 98.45% specificity, 96.78% sensitivity, 97.92% precision, 96.89% recall, and 97.93% F-score results in terms of DR classification than previous methods.
In this manuscript, Enhanced Elman Spike Neural Network (EESNN) optimized with red fox optimization algorithm is proposed for Sugarcane Yield Grade Prediction (SYGD-EESNN-RFOA). Initially, the sugar yield prediction d...
详细信息
In this manuscript, Enhanced Elman Spike Neural Network (EESNN) optimized with red fox optimization algorithm is proposed for Sugarcane Yield Grade Prediction (SYGD-EESNN-RFOA). Initially, the sugar yield prediction dataset is taken. Then the input data are pre-processed by hybrid decomposition method that is morphological filtering and extended empirical wavelet transformation (MF-EEWT) to retrieve the missing values. These pre-processed outputs are given to feature selection methods. During the process of feature selection, Entropy - Kurtosis-based feature selection method (EKBFS) is applied. These extracted features are fed to EESNN, and then it classifies the sugarcane yield as low grade, medium grade, and high grade. Generally, EESNN method does not indicate the use of any optimization strategies for calculating the best parameters to ensure accurate sugarcane yield forecast. To forecast the sugarcane production accurately, the red fox optimization algorithm (RFOA) is proposed. The proposed approach is carried out in Python;its performance is evaluated under performance metrics, such as precision, root mean square error, mean square error, mean absolute percentage error, convergence curve, and predicted percentage of changes in sugarcane yield during 2021-2027. The proposed SYGP-EESNN-RFOA framework attains higher accuracy of 27.5%, 16.65%, and 9.13%, 15.21% higher specificity compared with the existing methods.
Classification of electroencephalography (EEG) signals associated with Steady-state visually evoked potential (SSVEP) is prominent because of its potential in restoring the communication and controlling capability of ...
详细信息
Classification of electroencephalography (EEG) signals associated with Steady-state visually evoked potential (SSVEP) is prominent because of its potential in restoring the communication and controlling capability of paralytic people. However, SSVEP signals classification is a challenging task for researchers because of its low signal-to-noise ratio, non-stationary and high dimensional properties. A proficient technique has to be evolved to classify the SSVEP-based EEG data. In recent times, convolutional neural network (CNN) has reached a quantum leap in EEG signal classification. Therefore, the proposed system employs CNN to classify the SSVEP-based EEG signals. Though CNN has proved its proficiency in handling EEG signal classification problems, the calibration of hyperparameters is required to enhance the performance of the model. The calibration of a hyperparameter is a time-consuming task, hence proposed an automated hyperparameter optimization technique using the red fox optimization algorithm (RFO). The effectiveness of the algorithm is evaluated by comparing it with the performance of Harris Hawk optimization (HHO), Flower Pollination algorithm (FPA), Grey Wolf optimizationalgorithm (GWO) and Whale optimizationalgorithm (WOA) based hyperparameter optimized CNN applied to the SSVEP based EEG signals multiclass dataset. The experimental results infer that the proposed algorithm can achieve a testing accuracy of 88.91% which is higher than other comparative algorithms like HHO, FPA, GWO and WOA. The above-mentioned values clearly show that the proposed algorithm achieved competitive performance when compared to the other reported algorithm.
Municipal solid waste is considered to eliminate the problem of dumping and spreading in rural and urban areas of developing countries. Accumulation of solid wastes in open spaces receives greater concern in solid was...
详细信息
Municipal solid waste is considered to eliminate the problem of dumping and spreading in rural and urban areas of developing countries. Accumulation of solid wastes in open spaces receives greater concern in solid waste management systems because it leads to environmental hazards and health issues. To build a clean environment, it is essential to construct an advanced and intelligent waste management system to handle different compositions of waste materials. The significant step of waste management is the separation of waste components, which is normally carried out by manual operation. As a result, it can generate improper disposal of waste materials, to simplify the separation process mechanically, a novel automated Dense Net- BiLSTM-based redfox (DNBiLSTM-RF) approach is proposed in this paper. The proposed solid waste classification framework is analyzed by using waste data which is gathered from the Tehran waste management organization. The input waste data is preprocessed initially to transform raw amorphous data into appropriate data structures and extract the most significant dense and latent data features. The abnormal variations in waste patterns generate outliers which are effectively removed by applying the interquartile range (IQR) filtering process. Finally, the proposed DNBiLSTM-RF classifier accurately discriminates municipal waste materials into six different categories such as wood waste, textiles, food residues, rubber, paper, and plastics. The hyperparameters of the DenseNet-BiLSTM model are fine-tuned using a redfox (RF) optimizationalgorithm to enhance the classification performance of the model. The effectiveness of the DNBiLSTM-RF approach is evaluated using performance indicators namely root mean square error (RMSE), mean absolute error (MAE), the ratio of RMSE to the standard deviation (SD), Nash-Sutcliffe efficiency, coefficient of determination, recall, precision, F-measure, and accuracy. The analytic result demonstrates the feasibili
This study explores the use of a recent metaheuristic algorithm called a reptile search algorithm (RSA) to handle engineering design optimization problems. It is the first application of the RSA to engineering design ...
详细信息
This study explores the use of a recent metaheuristic algorithm called a reptile search algorithm (RSA) to handle engineering design optimization problems. It is the first application of the RSA to engineering design problems in literature. The RSA optimizer is first applied to the design of a bolted rim, which is constrained optimization. The developed algorithm is then used to solve the optimization problem of a vehicle suspension arm, which aims to solve the weight reduction under natural frequency constraints. As function evaluations are achieved by finite element analysis, the Kriging surrogate model is integrated into the RSA algorithm. It is revealed that the optimum result gives a 13% weight reduction compared to the original structure. This study shows that RSA is an efficient metaheuristic as other metaheuristics such as the mayfly optimizationalgorithm, battle royale optimizationalgorithm, multi-level cross-entropy optimizer, and red fox optimization algorithm.
Purpose This study aims to develop a trust mechanism in a Vehicular ad hoc Network (VANET) based on an optimized deep learning for selfish node detection. Design/methodology/approach The authors built a deep learning-...
详细信息
Purpose This study aims to develop a trust mechanism in a Vehicular ad hoc Network (VANET) based on an optimized deep learning for selfish node detection. Design/methodology/approach The authors built a deep learning-based optimized trust mechanism that removes malicious content generated by selfish VANET nodes. This deep learning-based optimized trust framework is the combination of the Deep Belief Network-based red fox optimization algorithm. A novel deep learning-based optimized model is developed to identify the type of vehicle in the non-line of sight (nLoS) condition. This authentication scheme satisfies both the security and privacy goals of the VANET environment. The message authenticity and integrity are verified using the vehicle location to determine the trust level. The location is verified via distance and time. It identifies whether the sender is in its actual location based on the time and distance. Findings A deep learning-based optimized Trust model is used to detect the obstacles that are present in both the line of sight and nLoS conditions to reduce the accident rate. While compared to the previous methods, the experimental results outperform better prediction results in terms of accuracy, precision, recall, computational cost and communication overhead. Practical implications The experiments are conducted using the Network Simulator Version 2 simulator and evaluated using different performance metrics including computational cost, accuracy, precision, recall and communication overhead with simple attack and opinion tampering attack. However, the proposed method provided better prediction results in terms of computational cost, accuracy, precision, recall, and communication overhead than other existing methods, such as K-nearest neighbor and Artificial Neural Network. Hence, the proposed method highly against the simple attack and opinion tampering attacks. Originality/value This paper proposed a deep learning-based optimized Trust framework for
暂无评论