Yazar "Alenezi, Fayadh" seçeneğine göre listele
Listeleniyor 1 - 20 / 24
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe Accurate 3D contrast-free myocardial infarction delineation using a 4D dual-stream spatiotemporal feature learning framework(Elsevier, 2023) Liu, Jinhao; Zhu, Xinglai; Xu, Chenchu; Xu, Lei; Polat, Kemal; Alenezi, FayadhAccurate 3D contrast-free myocardial infarction (MI) delineation has the potential to eliminate the need for toxic injections, thereby significantly advances diagnosis and treatment of MI. In this study, we propose a 4D dual-stream spatiotemporal feature learning framework (4D-DSS) that enables learning of 4D (3D + T) representation of the heart to accurately map the 3D MI regions, thereby directly delineating of 3D MI without contrast agent. This framework creatively introduces a dual-stream 3D spatiotemporal point cloud architecture enables to learn the myocardial 4D representation in both local and global aspects, and improve the comprehension and precision of the representation. Specifically, the framework utilizes the local spatiotemporal variation of individual point clouds to characterize minute distortions in myocardial regions and the global spatiotemporal variation of point cloud sequences to represent the overall myocardial motion between frames, thereby enables comprehensive learning of 3D myocardial motion and leverages these features to classify myocardial tissue into MI regions and normal regions. 4D-DSS significantly improved performance (with a precision increase of at least 4%) compared to four advanced methods. The results support the impact of our 4D-DSS framework on the development and implementation of 3D contrast-free myocardial infarction region delineation technology.& COPY; 2023 Elsevier B.V. All rights reserved.Öğe Arrhythmia diagnosis from ECG signal pulses with one-dimensional convolutional neural networks(Elsevier, 2023) Senturk, Umit; Polat, Kemal; Yucedag, Ibrahim; Alenezi, FayadhA noninvasive technique, electrocardiography (ECG), is crucial in the detection and treatment of cardiovascular disorders. Periodic beats make up ECG signals, and these beats vary based on the internal dynamics of the cardiovascular system. It is highly challenging to categorize ECG beats that can be understood by specialists in the area. In recent years, attempts have been made to use artificial intelligence programs in conjunction with a database of specified ECG beats to identify autonomous cardiovascular illness. In this investigation, the PTB Diagnostic ECG Database and the MIT-BIH Arrhythmia Database were used to attempt to classify arrhythmias. A one-dimensional convolutional neural network (CNN, or ConvNet) model was used to estimate the arrhythmia classes. The ECG beats defined in the database are divided into five classes: normal (N), supraventricular premature (S), premature ventricular contraction (V), ventricular and normal fusion (F), and Unclassifiable beats (U). The utilized one-dimensional convolutional neural network (1D-CNN VGG16) model’s average accuracy in classifying arrhythmias was found to be 99.12%. With the aid of this study, a system of experts has been built to assist specialized doctors in the healthcare system. The high estimating success of the used model will help in combining the right diagnosis with the right therapy and saving lives. © 2023 Elsevier Inc. All rights reserved.Öğe Attention based CNN model for fire detection and localization in real-world images(PERGAMON-ELSEVIER SCIENCE LTD, 2022) Majid, Saima; Alenezi, Fayadh; Masood, Sarfaraz; Ahmad, Musheer; Gündüz, Emine Selda; Polat, KemalFire is a severe natural calamity that causes significant harm to human lives and the environment. Recent works have proposed the use of computer vision for developing a cost-effective automated fire detection system. This paper presents a custom framework for detecting fire using transfer learning with state-of-the-art CNNs trained over real-world fire breakout images. The framework also uses the Grad-CAM method for the visualization and localization of fire in the images. The model also uses an attention mechanism that has significantly assisted the network in achieving better performances. It was observed through Grad-CAM results that the proposed use of attention led the model towards better localization of fire in the images. Among the plethora of models explored, the EfficientNetB0 emerged as the best-suited network choice for the problem. For the selected real-world fire image dataset, a test accuracy of 95.40% strongly supports the model's efficiency in detecting fire from the presented image samples. Also, a very high recall of 97.61 highlights that the model has negligible false negatives, suggesting the network to be reliable for fire detection.Öğe Automatic arrhythmia detection based on the probabilistic neural network with FPGA implementation(Hindawi Ltd, 2022) Srivastava, Rohini; Kumar, Basant; Alenezi, Fayadh; Alhudhaif, Adi; Althubiti, Sara A.; Polat, KemalThis paper presents a prototype implementation of arrhythmia classification using Probabilistic neural network (PNN). Arrhythmia is an irregular heartbeat, resulting in severe heart problems if not diagnosed early. Therefore, accurate and robust arrhythmia classification is a vital task for cardiac patients. The classification of ECG has been performed using PNN into eight ECG classes using a unique combination of six ECG features: heart rate, spectral entropy, and 4th order of autoregressive coefficients. In addition, FPGA implementation has been proposed to prototype the complete system of arrhythmia classification. Artix-7 board has been used for the FPGA implementation for easy and fast execution of the proposed arrhythmia classification. As a result, the average accuracy for ECG classification is found to be 98.27%, and the time consumed in the classification is found to be 17 seconds.Öğe Brain tumor classification using the modified ResNet50 model based on transfer learning(Elsevier Sci LTD, 2023) Sharma, Arpit Kumar; Nandal, Amita; Dhaka, Arvind; Zhou, Liang; Alenezi, Fayadh; Polat, KemalBrain tumour classification is essential for determining the type and grade and deciding on therapy appropriately. Several diagnostic methods are used in the therapeutic therapy to identify brain tumours. MRI, on the other hand, offers superior picture clarity, which is why specialists depend on it. Furthermore, detecting cancer through the manual division of brain tumours is a time-consuming, exhausting, and difficult job. The handdesigned outlines for planned brain tumour growth methods are present in the majority of the instances. Segmentation is a highly reliable and precise method for assessing therapy prognosis, planning, and outcomes. Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) advancements have enabled us to investigate the illness with high precision in a short period of time. Such technologies have produced some remarkable results, particularly in the last twenty years. Such breakthroughs provide doctors with the ability to evaluate the human anatomy using high-resolution sections. The most recent approaches can improve diagnostic precision when examining patients using non-invasive means. This work introduces a brain tumour detection method. The model grows using ResNet50, feature extraction, and augmentation. CNN's pre-trained datasets are used to fine-tune transfer learning. The proposed design utilised elements of the ResNet50 model, removing the final layer and adding four additional layers to meet work conditions. This study uses the improved ResNet50 model to present a novel deep-learning approach based on a transfer learning technique for evaluating brain cancer categorisation accuracy. Performance metrics were used to evaluate the effectiveness of the proposed model, and the results were compared to those obtained using state-of-the-art methods.Öğe Consistency- and dependence-guided knowledge distillation for object detection in remote sensing images(Pergamon-Elsevier Science Ltd, 2023) Chen, Yixia; Lin, Mingwei; He, Zhu; Polat, Kemal; Alhudhaif, Adi; Alenezi, FayadhAs one of the challenging tasks in the remote sensing (RS), object detection has been successfully applied in many fields. Convolution neural network (CNN) has recently attracted extensive attention and is widely used in the natural image processing. Nevertheless, RS images have cluttered scenes compared with natural images. As a result, the existing detectors perform poorly in RS images, especially with the complicated backgrounds. Moreover, the detection inference time and model volume of detectors in RS images often go unrecognized. To address the above issues, this study proposes a novel method for object detection in RS images, which is called the consistency- and dependence-guided knowledge distillation (CDKD). To this end, the spatial- and channeloriented structure discriminative modules (SCSDM) are put forward to extract the discriminative spatial locations and channels to which the teacher model pays attention. SCSDM improves the feature representation of the student model by effectively eliminating the influence of noises and the complicated backgrounds. Then, the consistency and dependence of the features between the teacher model and the student model are constructed under the guidance of SCSDM. Experimental results over public datasets for RS images demonstrate that our CDKD method surpasses the state-of-the-art methods effectively. Most of all, on the RSOD dataset, our CDKD method achieves 92% mean average precision with 3.3 M model volume and 588.2 frames per second.Öğe An effective hashing method using W-Shaped contrastive loss for imbalanced datasets(Pergamon-Elsevier Science Ltd, 2022) Alenezi, Fayadh; Öztürk, Şaban; Armghan, Ammar; Polat, KemalThe extraction of informative features from medical images and the retrieving of similar images from data repositories is vital for clinical decision support systems. Unlike general tasks such as medical image classification and segmentation, retrieval is more reliable in terms of interpretability. However, this task is quite challenging due to the multimodal and imbalanced nature of medical images. Because traditional retrieval methods use hand-crafted feature extraction guided approximate hashing functions, they often have problems capturing the latent characteristics of images. Deep learning based retrieval methods can eliminate drawbacks of hand-crafted feature extraction methods. However, in order for a deep architecture to produce high performance, large-scale datasets containing labeled and balanced samples are required. Since most medical datasets do not have these properties, existing hashing methods are not powerful enough to model patterns in medical images, which have a similar general appearance but subtle differences. In this study, a novel W-shaped contrastive loss (W-SCL) is proposed for skin lesion image retrieval on a dataset whose visual difference between classes is relatively low. We considerably improve the traditional contrastive loss (CL) performance by including label information for very similar skin lesion images. We use two benchmark datasets consisting of general images and two benchmark skin lesion datasets to test the proposed W-SCL performance. In addition, experiments are carried out using various pre-trained CNN and shallow CNN architectures. These extensive experiments reveal that the proposed method improves the mean average precision (mAP) performance by approximately 7% for general image datasets and approximately 12% for skin lesion datasets.Öğe An ensemble learning approach for resampling forgery detection using Markov process(Elsevier, 2023) Mehta, Rachna; Kumar, Karan; Alhudhaif, Adi; Alenezi, Fayadh; Polat, KemalResampling is an extremely well-known technique performed for Image forgery detection. It includes the changes in the content of a picture in terms of rotation, stretching/zooming, and shrinking, to frame a forged picture that is a localized forgery in comparison to the original picture. With the wrong intention, resampling forgery has been increased day by day, and its negative impact has been increased in criminology, law enforcement, forensics, research etc. Accordingly, the interest in the algorithm of image resampling forgery detection is significantly developed in image forensics. In this paper, a novel image resampling forgery detection technique has been proposed. In the proposed technique, two types of Markov feature with spatial and Discrete Cosine Transform domains have been extracted to recognize the resampling operation. The spatial domain gives the information for the distribution of the pixels and DCT gives the edge information. Further, these Markov features are consolidated. Due to high dimensionality hard thresholding technique is used for reducing the dimensionality. Then, these Markov features are applied to the set of models of different classifiers. With the utilization of classifiers, weighted majority voting values have been calculated during the ensemble classification. Unlike the other techniques, these weighted voting boundaries have been consequently balanced during the training process until the best accuracy has been obtained. However, it is very difficult to get best accuracy so for getting best accuracy this research needs to do lots of iterations and trained the dataset. For the comparative study very few research has been found for this resampling forgery technique with different interpolation techniques and classifier. Still, comparison has been done with some latest research work. The comparative analysis shows that the proposed ensemble learning-based algorithm provides the best outcomes with the accuracy of 99.12% for bicubic, 98.89% for bilinear, and 98.23% for lanczos3 kernel with considerably less complexity and high speed in comparison to prior techniques which are using single support vector machine for classification. Moreover, the proposed algorithm also detects a very low probability of error of 0.44% and detects the type of interpolation kernel, size of the forgery, and the type of resampling, whether it is up sampling and down sampling, using Graphical User Interface which has not been detected previously with multiple forgery detection.& COPY; 2023 Elsevier B.V. All rights reserved.Öğe Ensemble learning framework with GLCM texture extraction for early detection of lung cancer on CT images(Hindawi Ltd, 2022) Althubiti, Sara A.; Paul, Sanchita; Mohanty, Rajanikanta; Mohanty, Sachi Nandan; Alenezi, Fayadh; Polat, KemalLung cancer has emerged as a major cause of death among all demographics worldwide, largely caused by a proliferation of smoking habits. However, early detection and diagnosis of lung cancer through technological improvements can save the lives of millions of individuals affected globally. Computerized tomography (CT) scan imaging is a proven and popular technique in the medical field, but diagnosing cancer with only CT scans is a difficult task even for doctors and experts. This is why computer-assisted diagnosis has revolutionized disease diagnosis, especially cancer detection. This study looks at 20 CT scan images of lungs. In a preprocessing step, we chose the best filter to be applied to medical CT images between median, Gaussian, 2D convolution, and mean. From there, it was established that the median filter is the most appropriate. Next, we improved image contrast by applying adaptive histogram equalization. Finally, the preprocessed image with better quality is subjected to two optimization algorithms, fuzzy c-means and k-means clustering. The performance of these algorithms was then compared. Fuzzy c-means showed the highest accuracy of 98%. The feature was extracted using Gray Level Cooccurrence Matrix (GLCM). In classification, a comparison between three algorithms-bagging, gradient boosting, and ensemble (SVM, MLPNN, DT, logistic regression, and KNN)-was performed. Gradient boosting performed the best among these three, having an accuracy of 90.9%.Öğe Exposing low-quality deepfake videos of Social Network Service using Spatial Restored Detection Framework(Pergamon-Elsevier Science Ltd, 2023) Li, Ying; Bian, Shan; Wang, Chuntao; Polat, Kemal; Alhudhaif, Adi; Alenezi, FayadhThe increasing abuse of facial manipulation methods, such as FaceSwap, Deepfakes etc., seriously threatens the authenticity of digital images/videos on the Internet. Therefore, it is of great importance to identify the facial videos to confirm the contents and avoid fake news or rumors. Many researchers have paid great attention to the detection of deepfakes and put forward a number of deep-learning-based detection models. The existing approaches mostly face the performance degradation in detecting low-quality(LQ) videos, i.e. heavily compressed or low-resolution videos through some SNS (Social Network Service), resulting in the limitation in real applications. To address this issue, in this paper, a novel Spatial Restore Detection Framework(SRDF) is proposed for improving the detection performance for LQ videos by restoring spatial features. We designed a feature extraction-enhancement block and a mapping block inspired by super-resolution methods, to restore and enhance texture features. An attention module was introduced to guide the texture features restoration and enhancement stage attending to different local areas and restoring the texture features. Besides, an improved isolated loss was put forward to prevent the expansion of a single area concerned. Moreover, we adopted a regional data augmentation strategy to prompt feature restore and enhancement in the region attended. Extensive experiments conducted on two deepfake datasets have validated the superiority of the proposed method compared to the state-of-the-art, especially in the scenarios of detecting low-quality deepfake videos.Öğe Forecasting Covid-19 pandemic using prophet, ARIMA, and hybrid stacked LSTM-GRU models in India(Hindawi, 2022) Sah, Sweeti; Surendiran, B.; Dhanalakshmi, R.; Mohanty, Sachi Nandan; Alenezi, Fayadh; Polat, KemalDue to the proliferation of COVID-19, the world is in a terrible condition and human life is at risk. The SARS-CoV-2 virus had a significant impact on public health, social issues, and financial issues. Thousands of individuals are infected on a regular basis in India, which is one of the populations most seriously impacted by the pandemic. Despite modern medical and technical technology, predicting the spread of the virus has been extremely difficult. Predictive models have been used by health systems such as hospitals, to get insight into the influence of COVID-19 on outbreaks and possible resources, by minimizing the dangers of transmission. As a result, the main focus of this research is on building a COVID-19 predictive analytic technique. In the Indian dataset, Prophet, ARIMA, and stacked LSTM-GRU models were employed to forecast the number of confirmed and active cases. State-of-the-art models such as the recurrent neural network (RNN), gated recurrent unit (GRU), long short-term memory (LSTM), linear regression, polynomial regression, autoregressive integrated moving average (ARIMA), and Prophet were used to compare the outcomes of the prediction. After predictive research, the stacked LSTM-GRU model forecast was found to be more consistent than existing models, with better prediction results. Although the stacked model necessitates a large dataset for training, it aids in creating a higher level of abstraction in the final results and the maximization of the model's memory size. The GRU, on the other hand, assists in vanishing gradient resolution. The study findings reveal that the proposed stacked LSTM and GRU model outperforms all other models in terms of R square and RMSE and that the coupled stacked LSTM and GRU model outperforms all other models in terms of R square and RMSE. This forecasting aids in determining the future transmission paths of the virus.Öğe Graph-based link prediction between human phenotypes and genes(Hindawi Ltd, 2022) Patel, Rushabh; Guo, Yanhui; Alhudhaif, Adi; Alenezi, Fayadh; Althubiti, Sara A.; Polat, KemalDeep phenotyping is defined as learning about genotype-phenotype associations and the history of human illness by analyzing phenotypic anomalies. It is significant to investigate the association between phenotype and genotype. Machine learning approaches are good at predicting the associations between abnormal human phenotypes and genes. A novel framework based on machine learning is proposed to estimate the links between human phenotype ontology (HPO) and genes. The Orphanet's annotation parses the human phenotype-gene associations. An algorithm node2vec generates the embeddings for the nodes (HPO and genes). It performs node sampling on the graph using random walks and learns features on these sampled nodes for embedding. These embeddings were used downstream to predict the link between these nodes by supervised classifiers. Results show the gradient boosting decision tree model (LightGBM) has achieved an optimal AUROC of 0.904 and an AUCPR of 0.784, an optimal weighted F1 score of 0.87. LightGBM can detect more accurate interactions and links between human phenotypes and gene pairs.Öğe HOG transformation based feature extraction framework in modified Resnet50 model for brain tumor detection(Elsevier Science Ltd, 2023) Sharma, Arpit Kumar; Nandal, Amita; Dhaka, Arvind; Polat, Kemal; Alwadie, Raghad; Alenezi, FayadhBrain tumor happens due to the instant and uncontrolled cell growth. It may lead to death if not cured at an early stage. In spite of several promising results and substantial efforts in this research area, the real challenge is to provide the accurate classification and segmentation. The key issue in brain tumor detection develops from the irregular changes in the tumor size, shape and location. In assessing the MRI images, computer-aided diagnoses are playing an extraordinary role and can help clinicians/radiologist. Nowadays, brain tumor has become the most incursive ailment that leads to a very short life expectancy when it reaches its highest grade. This research paper has created a new model using histogram of gradient (HOG) based neural features from MRI images for tumors detection. This research has conducted the feature optimization approach to achieve additional instinctive features from the complex feature vector. We developed a Modified ResNet50 model with HOG technique. The modified ResNet50 model can accurately extract the deep feature using deep learning approach. This model is applied along with the upgraded layered architecture in order to keep the optimal computational efficiency. We have also used the augmentation and feature extraction techniques using machine learning-based ensemble classifier that further provides the optimized fusion vector to identify the tumor. Such hybrid approach provides excellent performance with the detection accuracy of 88% with HOG and modified ResNet50. The results are also compared with the recent state of art methods.Öğe A multi-stage melanoma recognition framework with deep residual neural network and hyperparameter optimization-based decision support in dermoscopy images(Pergamon-Elsevier Science Ltd, 2023) Alenezi, Fayadh; Armghan, Ammar; Polat, KemalThis paper developed a novel melanoma diagnosis model from dermoscopy images using a novel hybrid model. Melanoma is the most dangerous and rarest type of skin cancer. It is seen because of the uncontrolled prolif-eration of melanocyte cells that give color to the skin. Dermoscopy is a critical auxiliary diagnostic method in the differentiation of pigmented moles, which show moles by magnifying 10-20 times from skin cancers. This paper proposes a multi-stage melanoma recognition framework with skin lesion images obtained from dermoscopy. This model developed a practical pre-processing approach that includes dilation and pooling layers to remove hair details and reveal details in dermoscopy images. A deep residual neural network was then utilized as the feature extractor for processed images.Additionally, the Relief algorithm selected practical and distinctive features from these features. Finally, these selected features were fed to the input of the support vector machine (SVM) classifier. In addition, the Bayesian optimization algorithm was used for the optimum parameter selection of the SVM method. The International Skin Imaging Collaboration (ISIC-2019 and ISIC-2020) datasets were used to test the performance of the pro-posed model. As a result, the proposed model produced approximately 99% accuracy for classifying melanoma or benign from skin lesion images. These results show that the proposed model can help physicians to automatically identify melanoma based on dermatological imaging.Öğe A novel automatic audiometric system design based on machine learning methods using the brain's electrical activity signals(MDPI, 2023) Küçükakarsu, Mustafa; Kavsaoğlu, Ahmet Reşit; Alenezi, Fayadh; Alhudhaif, Adi; Alwadie, Raghad; Polat, KemalThis study uses machine learning to perform the hearing test (audiometry) processes autonomously with EEG signals. Sounds with different amplitudes and wavelengths given to the person tested in standard hearing tests are assigned randomly with the interface designed with MATLAB GUI. The person stated that he heard the random size sounds he listened to with headphones but did not take action if he did not hear them. Simultaneously, EEG (electro-encephalography) signals were followed, and the waves created in the brain by the sounds that the person attended and did not hear were recorded. EEG data generated at the end of the test were pre-processed, and then feature extraction was performed. The heard and unheard information received from the MATLAB interface was combined with the EEG signals, and it was determined which sounds the person heard and which they did not hear. During the waiting period between the sounds given via the interface, no sound was given to the person. Therefore, these times are marked as not heard in EEG signals. In this study, brain signals were measured with Brain Products Vamp 16 EEG device, and then EEG raw data were created using the Brain Vision Recorder program and MATLAB. After the data set was created from the signal data produced by the heard and unheard sounds in the brain, machine learning processes were carried out with the PYTHON programming language. The raw data created with MATLAB was taken with the Python programming language, and after the pre-processing steps were completed, machine learning methods were applied to the classification algorithms. Each raw EEG data has been detected by the Count Vectorizer method. The importance of each EEG signal in all EEG data has been calculated using the TF-IDF (Term Frequency-Inverse Document Frequency) method. The obtained dataset has been classified according to whether people can hear the sound. Naive Bayes, Light Gradient Strengthening Machine (LGBM), support vector machine (SVM), decision tree, k-NN, logistic regression, and random forest classifier algorithms have been applied in the analysis. The algorithms selected in our study were preferred because they showed superior performance in ML and succeeded in analyzing EEG signals. Selected classification algorithms also have features of being used online. Naive Bayes, Light Gradient Strengthening Machine (LGBM), support vector machine (SVM), decision tree, k-NN, logistic regression, and random forest classifier algorithms were used. In the analysis of EEG signals, Light Gradient Strengthening Machine (LGBM) was obtained as the best method. It was determined that the most successful algorithm in prediction was the prediction of the LGBM classification algorithm, with a success rate of 84%. This study has revealed that hearing tests can also be performed using brain waves detected by an EEG device. Although a completely independent hearing test can be created, an audiologist or doctor may be needed to evaluate the results.Öğe A novel facial emotion recognition method for stress inference of facial nerve paralysis patients(Pergamon-Elsevier Science Ltd, 2022) Xu, Cuiting; Yan, Chunchuan; Jiang, Mingzhe; Alenezi, Fayadh; Alhudhaif, Adi; Polat, KemalFacial nerve paralysis results in muscle weakness or complete paralysis on one side of the face. Patients suffer from difficulties in speech, mastication and emotional expression, impacting their quality of life by causing anxiety and depression. The emotional well-being of a facial nerve paralysis patient is usually followed up during and after treatment as part of quality-of-life measures through questionnaires. The commonly used questionnaire may help recognize whether a patient has been through a depressive state but is unable to understand their basic emotions dynamically. Automatic emotion recognition from facial expression images could be a solution to help understand facial nerve paralysis patients, recognize their stress in advance, and assist their treatment. However, their facial expressions are different from healthy people due to facial muscle inability, which makes existing emotion recognition data and models from healthy people invalid. Recent studies on facial images mainly focus on the automatic diagnosis of facial nerve paralysis level and thus lack full basic emotions. Different nerve paralysis levels also increase inconsistency in expressing the same emotion among patients. To enable emotion recognition and stress inference from facial images for facial nerve paralysis patients, we established an emotional facial expressions dataset from 45 patients with six basic emotions. The problem of limited data size in building a deep learning model VGGNet was solved by leveraging facial images from healthy people in transfer learning. Our proposed model reached an accuracy of 66.58% recognizing basic emotions from patients, which was 19.63% higher than the model trained only from the facial nerve paralysis data and was 42.69% higher than testing directly on the model trained from healthy data. Logically, the results show that patients with less severe facial nerve paralysis reached a higher emotion recognition accuracy. Additionally, although disgust, anger, and fear were especially challenging to specify from each other, the accuracy was 85.97% recognizing any stress-related negative emotions, making stress inference feasible.Öğe A novel multi-task learning network based on melanoma segmentation and classification with skin lesion images(MDPI, 2023) Alenezi, Fayadh; Armghan, Ammar; Polat, KemalMelanoma is known worldwide as a malignant tumor and the fastest-growing skin cancer type. It is a very life-threatening disease with a high mortality rate. Automatic melanoma detection improves the early detection of the disease and the survival rate. In accordance with this purpose, we presented a multi-task learning approach based on melanoma recognition with dermoscopy images. Firstly, an effective pre-processing approach based on max pooling, contrast, and shape filters is used to eliminate hair details and to perform image enhancement operations. Next, the lesion region was segmented with a VGGNet model-based FCN Layer architecture using enhanced images. Later, a cropping process was performed for the detected lesions. Then, the cropped images were converted to the input size of the classifier model using the very deep super-resolution neural network approach, and the decrease in image resolution was minimized. Finally, a deep learning network approach based on pre-trained convolutional neural networks was developed for melanoma classification. We used the International Skin Imaging Collaboration, a publicly available dermoscopic skin lesion dataset in experimental studies. While the performance measures of accuracy, specificity, precision, and sensitivity, obtained for segmentation of the lesion region, were produced at rates of 96.99%, 92.53%, 97.65%, and 98.41%, respectively, the performance measures achieved rates for classification of 97.73%, 99.83%, 99.83%, and 95.67%, respectively.Öğe Performance evaluation of deep e-CNN with integrated spatial-spectral features in hyperspectral image classification(Elsevier Science Ltd, 2022) M., Kavitha; Gayathri, R.; Polat, Kemal; Alhudhaif, Adi; Alenezi, FayadhDeep neural networks, an emerging paradigm in deep learning, have proven to make feature extraction from remote sensing data easier. Deep learning has been shown to be capable of effectively classifying hyperspectral images (HSI). Deep convolutional neural networks (CNNs) are one of the most effective approaches for HSI classification. The deep learning architecture needs to be capable of providing a better spatial-spectral classification performance. In increasing the depth of layers in CNN might lead to overfitting issues. As spatial-spectral information is not correlated along different layers, hence information is lost. This paper attempts to solve these problems by presenting an enhanced-CNN. Initially, proposed e-CNN method explores the merging of the outputs of successive two layers within the huge convolutional block and the merged feature extract outcome is fed as the input to the next layer, which renders relevant feature extraction. Then, from low-level layers to deep high-level layers, spectral-spatial features are retrieved by concatenating the spectral features to four-stage spatial features. Finally, to communicate with hybridized extracted feature information, a 1 x 1 convolution layer is used throughout the block. With the limited training samples and the provided pixel-size the proposed e-CNN model works much effectively. In order to obtain the standard generalizing ability of classification an adaptive AdaBound optimization method is used. Finally, HSI classification is performed with the enhanced CNN model. The existing models and optimizers (SGD, AdaGrad, AdaDelta, Adam) are used to compare the results. The experiments were carried out on widely used HSI datasets (i.e., Indian Pines and Salinas) and the result of the proposed e-CNN model with AdaBound optimizer obtains ~2% higher accuracy compared with existing methods. Optimization result of e-CNN model with AdaBound optimizer have the highest classification accuracy in the least amount of time.Öğe Robust underwater image enhancement with cascaded multi-level sub-networks and triple attention mechanism(Pergamon-Elsevier Science Ltd, 2024) Zhang, Dehuan; Wu, Chenyu; Zhou, Jingchun; Zhang, Weishi; Lin, Zifan; Polat, Kemal; Alenezi, FayadhWith the growing exploration of marine resources, underwater image enhancement has gained significant attention. Recent advances in convolutional neural networks (CNN) have greatly impacted underwater image enhancement techniques. However, conventional CNN-based methods typically employ a single network structure, which may compromise robustness in challenging conditions. Additionally, commonly used UNet networks generally force fusion from low to high resolution for each layer, leading to inaccurate contextual information encoding. To address these issues, we propose a novel network called Cascaded Network with Multi-level Sub-networks (CNMS), which encompasses the following key components: (a) a cascade mechanism based on local modules and global networks for extracting feature representations with richer semantics and enhanced spatial precision, (b) information exchange between different resolution streams, and (c) a triple attention module for extracting attention-based features. CNMS selectively cascades multiple subnetworks through triple attention modules to extract distinct features from underwater images, bolstering the network's robustness and improving generalization capabilities. Within the sub-network, we introduce a Multi-level Sub-network (MSN) that spans multiple resolution streams, combining contextual information from various scales while preserving the original underwater images' high-resolution spatial details. Comprehensive experiments on multiple underwater datasets demonstrate that CNMS outperforms state-of-the-art methods in image enhancement tasks.Öğe Smooth quantile regression and distributed inference for non-randomly stored big data(Pergamon-Elsevier Science Ltd, 2023) Wang, Kangning; Jia, Jiaojiao; Polat, Kemal; Sun, Xiaofei; Alhudhaif, Adi; Alenezi, FayadhIn recent years, many distributed algorithms towards big data quantile regression have been proposed. However, they all rely on the data are stored in random manner. This is seldom in practice, and the violation of this assumption can seriously degrade their performance. Moreover, the non-smooth quantile loss brings inconvenience in both computation and theory. To solve these issues, we first propose a convex and smooth quantile loss, which converges to the quantile loss uniformly. Then a novel pilot sample surrogate smooth quantile loss is constructed, which can realize communication-efficient distributed quantile regression, and overcomes the non-randomly distributed nature of big data. In theory, the estimation consistency and asymptotic normality of the resulting distributed estimator are established. The theoretical results guarantee that the new method is adaptive to the situation where the data are stored in any arbitrary way, and can work well just as all the data were pooled on a single machine. Numerical experiments on both synthetic and real data verify the good performance of the new method.
