AlexNet Network Research Articles

Transfer-learning based skin cancer diagnosis using fine-tuned AlexNet by Marine Predators Algorithm

<span>Melanoma represents one of the most dangerous manifestations of skin cancer. According to statistics, 55% of patients with skin cancer have lost their lives as a result of this disease. Early diagnosis of this condition will significantly reduce mortality rates and save lives. In recent years, deep learning methods have shown promising results and captured the attention of researchers in this field. One common approach is the use of pre-trained deep neural networks. In this work, a pre-trained AlexNet networks, which are networks with specified architecture and weights is used to automatic skin melanoma diagnosis. In the transfer learning phase, by reducing the learning rate, the pre-trained network is trained to recognize Skin cancer, which is called fine-tuning. In addition, Hyperparameters of the AlexNet network have been optimized by the Marine Predators Algorithm (MPA) to enhance the network performance. Experimental findings show the satisfactory efficiency of the presented approach, with an accuracy rate of 98.47%. The outcomes demonstrate the effectiveness of the suggested approach in contrast to alternative existing methods.</span>

Efficient Layer Optimizations for Deep Neural Networks

Deep neural networks (DNNs) have technical issues such as long training time as the network size increases. Parameters require significant memory, which may cause migration issues for embedded devices. DNNs applied various pruning techniques to reduce the network size in deep neural networks, but many problems still exist when applying the pruning techniques. Among neural networks, several applications applied autoencoders for reconstruction and dimension reduction. However, network size is a disadvantage of autoencoders since the architecture of the autoencoders has a double workload due to the encoding and decoding processes. In this research, we chose autoencoders and two deep neural networks AlexNet and VGG16 to apply out of order layer pruning. We perform the sensitivity analysis to explore the performance variations for the network architecture and network complexity through an out of order layer pruning mechanism. As a result of applying the proposed layer pruning scheme to the autoencoder, we developed the accordion autoencoder (A2E) and applied credit card fraud detection and MNIST classification. Our results show 4.9 Percent and 13.6 Percent performance drops, respectively, but we observe a significant reduction in network complexity, 85.1 Percent and 94.5 Percent for each application. We extend the out of order layer pruning to deeper learning networks. In our approach, we propose a simple yet efficient scheme, accuracy aware structured filter pruning based on the characterization of each convolutional layer combined with the quantization of fully connected layers. We investigate the accuracy and compression rate of each layer using a fixed pruning ratio, and then the pruning priority is rearranged depending on the accuracy of each layer. Our analysis of layer characterization shows that the pruning order of the layers does affect the final accuracy of the deep neural network. Based on our experiments using the proposed pruning scheme, the parameter size in the AlexNet can be up to 47.28x smaller than the original model. We also obtained comparable results for VGG16, achieving a maximum compression rate of 35.21x.

Medical Image Zero Watermarking Algorithm based on dual-tree complex wavelet transform, AlexNet and discrete cosine transform

This paper proposed for medical image using zero watermarking, dual-tree complex wavelet transform (DTCWT)-AlexNet and discrete cosine transform (DCT). Furthermore, we utilize the pre-training network AlexNet to extract significant characteristics from medical images. In addition, DTCWT and DCT are employed to convert the complex features, while a perceptual hash function is utilized to cause the feature vector. The original watermark is chaotically scrambled, an exclusive-OR (XOR) gate is inserted into the medical image, and a logical key vector is generated and stored to encrypt the watermark. In general, the medical image under test is used to cutting the important features and create a feature vector using the feature extraction technique. Subsequently, the logical key vector and the feature vector undergo an XOR operation to produce the encrypted watermark. By calculating the normalized correlation (NC) coefficient, the recovered watermark can be utilized to regulate the rights and specifics of the watermark in the medical image after obtaining and decrypting the encrypted watermark with greater than 0.5 NC values. The algorithm's design method tackles the challenge of protecting watermarks from conventional (Gaussian noise, JPEG, median filter) and geometric attacks ( rotation clockwise, anticlockwise, scaling, translation left, translation right, clipping X and Y direction) by incorporating the principles of DTCWT, AlexNet, DCT, and zero watermarking. Encrypting the watermark image through a scrambling process enhances the security of the approach, by ensuring that the watermark should not affect the quality of the image in any way. The research demonstrates that the proposed algorithm is robust and secure for protecting sensitive health information.

Classification and removal of hazy images based on a transmission fusion strategy using the Alexnet network

Classification and removal of hazy images based on a transmission fusion strategy using the Alexnet network

Deep transfer learning architecture for suspension system fault diagnosis using spectrogram image and CNN

The suspension system plays a critical role in automobiles, ensuring the safety and comfort of vehicle occupants. However, extended usage, varying road conditions, external forces, and heavy loads can result in damage and faults within the internal components of the suspension system. To mitigate the occurrence of suspension system failures, the development of an effective fault diagnosis system for suspension components becomes imperative. Traditional fault diagnosis techniques often heavily rely on human expertise, which comes with certain limitations. In response, researchers have embraced intelligent fault diagnosis techniques, with transfer learning-based fault diagnosis emerging as a highly effective approach. By leveraging transfer learning, it becomes possible to extract and select fault-specific features for classification purposes. Deep learning-based methods, with their capacity to extract significant features and essential information from raw data, offer notable advantages. Despite these advantages, the implementation of deep learning-based fault diagnosis in suspension systems remains relatively unexplored and limited. In this article, a deep transfer learning architecture specifically designed for fault diagnosis in suspension systems is proposed. The approach involves employing 12 pre-trained networks and tuning them to identify the optimal model for fault diagnosis. Time domain vibration signals obtained from suspension systems under seven fault conditions and one good condition are transformed into spectrogram images. These images are then pre-processed and used as input for the pre-trained networks in fault classification. The results demonstrate that among the 12 pre-trained networks, AlexNet outperforms the others in terms of classification accuracy while requiring the least amount of training time. Therefore, AlexNet network in conjunction with the spectrogram images of time domain vibration signals for applications in suspension system fault diagnosis is highly recommend.

Research on Input Schemes for Polarimetric SAR Classification Using Deep Learning

This study employs the reflection symmetry decomposition (RSD) method to extract polarization scattering features from ground object images, aiming to determine the optimal data input scheme for deep learning networks in polarimetric synthetic aperture radar classification. Eight distinct polarizing feature combinations were designed, and the classification accuracy of various approaches was evaluated using the classic convolutional neural networks (CNNs) AlexNet and VGG16. The findings reveal that the commonly employed six-parameter input scheme, favored by many researchers, lacks the comprehensive utilization of polarization information and warrants attention. Intriguingly, leveraging the complete nine-parameter input scheme based on the polarization coherence matrix results in improved classification accuracy. Furthermore, the input scheme incorporating all 21 parameters from the RSD and polarization coherence matrix notably enhances overall accuracy and the Kappa coefficient compared to the other seven schemes. This comprehensive approach maximizes the utilization of polarization scattering information from ground objects, emerging as the most effective CNN input data scheme in this study. Additionally, the classification performance using the second and third component total power values (P2 and P3) from the RSD surpasses the approach utilizing surface scattering power value (PS) and secondary scattering power value (PD) from the same decomposition.

A Mighty Image Retrieval Descriptor Based on Machine Learning and Gaussian Derivative Filter

The development of new image descriptor has always been an important topic to improve the efficiency of content- based image classification and retrieval. Improvements and developments in machine learning and deep learning algorithms as well as artificial intelligence algorithms are widely used by researchers to obtain effective CBIR descriptors. In our article, we will present a robust image descriptor, extended by machine learning and deep learning algorithms. The descriptor is provided through a Gaussian derivative filter scaffold named GDF-HOG with an enhanced convolutional neural network (CNN) AlexNet, to reduce the dimensions we used the principal component analysis algorithm. The experimental results were carried out on Oliva and Torralba, Caltech-101, Wang and Coil100 datasets. Experiments show that the accuracy of the proposed method is 98.23% for Coil-100%, 95.92% for Corel-1000, value 87.17 and 94.6% for Oliva and Torralba. In comparison our results with other descriptor image classifiers show that they achieved accuracy increases of 0.12% on average and up to 3.23%. These experimental results affirm the advantage of the proposed descriptor over existing systems based in terms of average accuracy. the proposed descriptor improves the precision, and also reduces the complexity of the calculation.

Research on the Vanishing Point Detection Method Based on an Improved Lightweight AlexNet Network for Narrow Waterway Scenarios

When an unmanned surface vehicle (USV) navigates in narrow waterway scenarios, its ability to detect vanishing points accurately and quickly is highly important for safeguarding its navigation safety and realizing automated navigation. We propose a novel approach for detecting vanishing points based on an improved lightweight AlexNet. First, a similarity evaluation calculation method based on image texture features is proposed, by which some scenarios are selected from the filtered Google Street Road Dataset (GSRD). These filtered scenarios, together with the USV Inland Dataset (USVID), compose the training dataset, which is manually labeled according to a non-uniformly distributed grid level. Next, the classical AlexNet was adjusted and optimized by constructing sequential connections of four convolutional layers and four pooling layers and incorporating the Inception A and Inception C structures in the first two convolutional layers. During model training, we formulate vanishing point detection as a classification problem using an output layer with 225 discrete possible vanishing point locations. Finally, we compare and analyze the labeled vanishing point with the detected vanishing point. The experimental results show that the accuracy of our method and the state-of-the-art algorithmic vanishing point detector improves, indicating that our improved lightweight AlexNet can be applied in narrow waterway navigation scenarios and can provide a technical reference for autonomous navigation of USVs.

A Simple Model for Fault Detection of UAV Photoelectric Load Based on Image Classification

A simple image-based unmanned aerial vehicle (UAV) optoelectronic load fault detection model is proposed to address the types of distortion that occur during the imaging process due to device failures. The model adopts the classic AlexNet network, adjusts the key parameters of the network, adds a BN layer, and then optimizes the original network using LeakyReLU, achieving a fault detection model based on image classification. The experimental results show that the clas-sification accuracy of our model has reached over 90%, which is higher than the comparison method. Moreover, the training samples required for our model are small and easy to implement.

Clinical Medical Detection and Diagnosis Technology Based on the AlexNet Network Model

With the rapid development of deep learning technology, its application in the medical field is more and more extensive. This paper mainly discusses the clinical medical detection and diagnosis technology based on AlexNet network model. This paper first introduces the basic principle and characteristics of AlexNet network model, then summarizes its application in the medical field, and then elaborates the implementation process of the technology, including data preprocessing, model training and optimization. Finally, the effectiveness and reliability of this technique are verified by experiments, and its future development is prospected.

Optimization of Nonlinear Convolutional Neural Networks based on Improved Chameleon Group Algorithm

In order to solve the most difficult problem of the architectural model established by CNN in solving specific problems, which results in parameter overflow and inefficient training, an optimization algorithm for nonlinear convolutional neural networks based on improved chameleon swarm algorithm is proposed. This article mainly introduces the use of Chameleon Swarm Optimization (PSO) algorithm to research the parameters of CNN architecture, solve them, and achieve the optimization of the optimization model.Although the number of parameters that need to be set up in CNN is very large, this method can find better testing space for Alexnet samples with 5 different images. In order to improve the performance of the improved pruning algorithms, two candidate pruning algorithms are also proposed. The experimental results show that compared with the traditional Alexnet model, the improved pruning method improves the image recognition ability of the Caffe primary parameter set from 1.3% to 5.7%. This method has wide applicability and can be applied to most neural networks which do not require any special functional modules of the Alexnet network model.

Accuracy study of image classification for reverse vending machine waste segregation using convolutional neural network

This study aims to create a sorting system with high accuracy that can classify various beverage containers based on types and separate them accordingly. This reverse vending machine (RVM) provides an image classification method and allows for recycling three types of beverage containers: drink carton boxes, polyethylene terephthalate (PET) bottles, and aluminium cans. The image classification method used in this project is transfer learning with convolutional neural networks (CNN). AlexNet, GoogLeNet, DenseNet201, InceptionResNetV2, InceptionV3, MobileNetV2, XceptionNet, ShuffleNet, ResNet 18, ResNet 50, and ResNet 101 are the neural networks that used in this project. This project will compare the F1-score and computational time among the eleven networks. The F1-score and computational time of image classification differs for each neural network. In this project, the AlexNet network gave the best F1-score, 97.50% with the shortest computational time, 2229.235 s among the eleven neural networks.

Diagnosis method of transformer winding mechanical deformation fault based on sliding correlation of FRA and series transfer learning

Frequency response analysis (FRA) is a common method for diagnosing transformer winding faults. There are many ways to interpret FRA, however, extracting fault features from FRA still faces challenges because the FRA signature of each transformer is different, which can make the generalization of FRA diagnosis difficult. Based on this, this paper proposes a diagnosis method based on FRA sliding correlation and series transfer learning. The study collected FRA data from three transformers, two obtained through simulation, while the third obtained through experiment. During the simulation, the synthetic minority over-sampling technique (smote) algorithm was used to expand the data set to solve the problem of unbalanced and insufficient data sets for different fault types. The FRA signature was processed by sliding correlation to extract fault features, and the FRA sequence data was transformed into a heat map with correlation coefficient values distinguished by color. Next, the AlexNet network was transferred twice using the simulation data, and the experimental data was used for the third transfer learning. The experimental results showed that the diagnosis method has high accuracy and strong generalization.

APPLICATION OF MODERN COMPUTER VISION ALGORITHMS TO MANAGE WITH THE COUNTING OF IMAGE OBJECTS

This article is devoted to the research of convolutional neural network architectures in counting objects in an image. Currently, methods using regression are gaining popularity. In this article in order to solve an object counting task using regression method, modifications of the reference convolutional neural networks AlexNet, VGG16, and ResNet 50, which were developed for image classification, were used. Modification presented by replacing the second part of the neural network, which classifies images, with one fully connected layer, consisting of one neuron without activating function. In experiments, modified architectures of the reference convolutional networks were initialized as folows: using random initialization of the weights and using pretrainedined weights trained on the ImageNet dataset. The results of experiments, which confirm the performance of the proposed models and the use of the neuroplasticity method to solve the problem using regression are preented. The database of images of bacterial cells was used as training and testing material.

Diagnosis of Fatty Liver Using a Hybrid Approach of Deep Learning and Ensemble Learning

Fatty liver disease is one of those health problems which, if not taken seriously, might lead to the most menacing complications. Henceforth, the diagnosing phase has to be both accurate and just-in-time to allow proper treatment. The approach based on the usage of the convolutional neural network AlexNet for feature extraction of a complex nature and decision-making by the AdaBoost classifier is hence introduced. Extremely high capability of feature extraction by AlexNet made its implementation reach a detection accuracy of 88.5%, in general 5% better than earlier practices. Results have proven that this approach can extract unique features from ultrasound images, which would be useful in diagnosing and managing fatty liver diseases. This could be instrumental in helping the physicians to be right on the spot in the accuracy and quality of diagnosis.

Artificial Neural Network-Powered, Driverless Vehicle Concept Development

Autonomous cars are now possible due to significant advances in robotics and intelligent control systems. Before these vehicles can safely operate in traffic and other hostile environments, there are many navigation, vision, and control issues. We want techniques that are both cost-effective and efficient, so that the field of research and academia may fully embrace self-driving cars. Within this scenario, we need something that can convert people to autonomous automobiles and include existing vehicles so that academics and explorers can access them. This study proposes a flexible mechanical layout that can be assembled in a short time and installed in most modern automobiles; it can also be used as a stepping stone in the development of autonomous vehicles. Using various actuators, conventional automobiles can be converted into autonomous vehicles. In the context of motor vehicle automation, motors are often used as actuators. In addition to motors, a pneumatic system was developed to automate the predetermined steps. An autonomous vehicle's mechanical arrangement is crucial, and it must be regularly updated and built to be robust in the face of dynamic conditions. We re-implemented two additional convolutional neural networks in an effort to conduct an objective test of their proposed network and compare our system's structure, technical complexity, and performance test during autonomous driving with theirs. This predicted network is around 250 times larger than the Alex Net network and four times larger than Pilot Net after training. Although the complexity and measurement of the publication's system are lower than other models that contribute lower latency and greater speed throughout inference, the operation was claimed by our system, which achieved autonomous driving with an equivalent efficacy as that achieved with two other models. The projected deep neural system reduces the need to infer ultra-fast computational hardware. This is important for cost efficiency, scale, and cost.

Application of AI intelligent vision detection technology using deep learning algorithm

This study aims to design efficient and reliable artificial intelligence vision detection models to improve detection efficiency and accuracy. The study filters defect-free images by image preprocessing and region of interest detection techniques. AlexNet network is enhanced by introducing attention mechanism modules, deep separable convolutions, and more to effectively boost the network's feature extraction capacity. An area convolutional neural network is developed to rapidly identify and locate defects on steel plate surfaces, utilizing an enhanced AlexNet network for feature extraction. Results demonstrated that the algorithm attained an average detection rate of 98 % and can identify defects in a minimal time of only 0.0011 seconds. For the detection of six types of steel plate defects, the average accuracy of the optimized fast regional convolutional neural network reached more than 0.9, especially for the detection of small-size defects with excellent performance. This improved AlexNet network has a great advantage in F1 value. The conclusion of the study shows that the designed artificial intelligence vision detection model has high detection accuracy, speed, and performance stability in steel plate surface defect detection and has a wide range of application prospects.

Predictive stroke risk model with vision transformer-based Doppler features.

Acute stroke is the leading cause of death and disability globally, with an estimated 16 million cases each year. The progression of carotid stenosis reduces blood flow to the intracranial vasculature, causing stroke. Early recognition of ischemic stroke is crucial for disease treatment and management. A computer-aided diagnosis (CAD) system was proposed in this study to rapidly evaluate ischemic stroke in carotid color Doppler (CCD). Based on the ground truth from the clinical examination report, the vision transformer (ViT) features extracted from all CCD images (513 stroke and 458 normal images) were combined in machine learning classifiers to generate the likelihood of ischemic stroke for each image. The pretrained weights from ImageNet reduced the time-consuming training process. The accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve were calculated to evaluate the stroke prediction model. The chi-square test, DeLongtest, and Bonferroni correction for multiple comparisons were applied to deal with the type-I error. Only p values equal to or less than 0.00125 were considered to be statistically significant. The proposed CAD system achieved an accuracy of 89%, a sensitivity of 94%, a specificity of 84%, and an area under the receiver operating characteristic curve of 0.95, outperforming the convolutional neural networks AlexNet (82%, p<0.001), Inception-v3 (78%, p<0.001), ResNet101 (84%, p<0.001), and DenseNet201 (85%, p<0.01). The computational time in model training was only 30 s, which would be efficient and practical in clinical use. The experiment shows the promising use of CCD images in stroke estimation. Using the pretrained ViT architecture, the image features can be automatically and efficiently generated without human intervention. The proposed CAD system provides a rapid and reliable suggestion for diagnosing ischemic stroke.

CPD-CCNN: classification of pepper disease using a concatenation of convolutional neural network models

Agricultural products are vital to the sustainability of the economies of developing countries. Most developing countries’ economies such as Ethiopia heavily rely on agriculture. On a global scale, the pepper crop is one of the most important agricultural products in terms of human food security. However, it is susceptible to a variety of diseases which include blight leaf disease, gray leaf spot, common rust, fruit rot disease, powdery mildew symptoms on pepper leaf, and other related diseases that are all common today. Currently, more than 34 different pepper diseases have been discovered, resulting in a 33% average yield loss in pepper cultivation. Conventionally, farmers detect the disease using visual observation but this has its own demerits as it is usually not accurate and usually time consuming. In the past, a number of researchers have presented various methods for classifying pepper plant disease, especially using image processing and deep learning techniques. However, earlier studies have shown that binary classification requires improvement as some classes were more challenging to identify than others. In this study, we propose a concatenated neural network of the extracted features of VGG16 and AlexNet networks to develop a pepper disease classification model using fully connected layers. The development of the proposed concatenated CNN model includes steps such as dataset collection, image preprocessing, noise removal, segmentation, feature extraction, and classification. Finally, the proposed concatenated CNN model was evaluated, providing a training classification accuracy of 100%, validation accuracy of 97.29%, and testing accuracy of 95.82%. In general, it can be concluded from the findings of the study that the proposed concatenated model is suitable for identifying pepper leaf and fruit diseases from digital images of pepper.

Prediction and Classification Lung Cancer using Machine Learning

In wealthy nations, lung cancer is one of the most lethal diseases, and early detection is challenging. One of the most difficult problems people have faced recently is lung cancer diagnosis and treatment. Every day, early tumor diagnosis will continue to save many lives all around the world. This research presents a strategy that integrates a Convolutional Neural Network (CNN) with the AlexNet Network Model to categorize lung cancers as benign or malignant. One of the transfer learning models is AlexNet CNN. Here, in this project, the aim will be focussed onto list, discuss, and analyse several ML algorithm to classify and detect lung cancer and its stages. The proposed CNN achieves a high degree of accuracy, which is more effective than the accuracy attained by traditional neural network systems. Keywords: Alex Net , Convolution Neural Network, Machine learning, , Transfer learning, Tumor.