Neural Network AlexNet Research Articles

Quasi-visualizable detection of deep sub-wavelength defects in patterned wafers by breaking the optical form birefringence

Abstract In integrated circuit (IC) manufacturing, fast, nondestructive, and precise detection of defects in patterned wafers, realized by bright-field microscopy, is one of the critical factors for ensuring the final performance and yields of chips. With the critical dimensions of IC nanostructures continuing to shrink, directly imaging or classifying deep-subwavelength defects by bright-field microscopy is challenging due to the well-known diffraction barrier, the weak scattering effect, and the faint correlation between the scattering cross-section and the defect morphology. Herein, we propose an optical far-field inspection method based on the form-birefringence scattering imaging of the defective nanostructure, which can identify and classify various defects without requiring optical super-resolution. The technique is built upon the principle of breaking the optical form birefringence of the original periodic nanostructures by the defect perturbation under the anisotropic illumination modes, such as the orthogonally polarized plane waves, then combined with the high-order difference of far-field images. We validated the feasibility and effectiveness of the proposed method in detecting deep subwavelength defects through rigid vector imaging modeling and optical detection experiments of various defective nanostructures based on polarization microscopy. On this basis, an intelligent classification algorithm for typical patterned defects based on a dual-channel AlexNet neural network has been proposed, stabilizing the classification accuracy of λ/16-sized defects with highly similar features at more than 90%. The strong classification capability of the two-channel network on typical patterned defects can be attributed to the high-order difference image and its transverse gradient being used as the network's input, which highlights the polarization modulation difference between different patterned defects more significantly than conventional bright-field microscopy results. This work will provide a new but easy-to-operate method for detecting and classifying deep-subwavelength defects in patterned wafers or photomasks, which thus endows current online inspection equipment with more missions in advanced IC manufacturing.

A bearing fault diagnosis approach based on an improved neural network combined with transfer learning

Abstract In modern industrial systems, bearing failures account for 30%–40% of industrial machinery faults. Traditional convolutional neural network suffers from gradient vanishing and overfitting, resulting in a poor diagnostic accuracy. To address the issues, a new bearing fault diagnosis approach was proposed based on an improved AlexNet neural network combined with transfer learning. After decomposition and noise-reduction, reconstructed vibration signals were transformed into 2D images, then input into the improved AlexNet for training and follow-up transfer learning. Program auto-tuning and image-enhancing techniques were employed to increase the diagnostic accuracy in this study. The approach was verified with the datasets from Case Western Reserve University (CWRU), Jiangnan University (JNU), and the Association for Mechanical Failure Prevention Technology (MFPT). The results showed that the diagnostic accuracies by normal learning were more than 97% for CWRU and JNU datasets, and 100% for MFPT dataset. After transfer learning, the accuracies all reached above 99.5%. The proposed approach was demonstrated to be able to effectively diagnose the bearing faults.

Diagnosis of Alzheimer's Disease and Mild Cognitive Impairment Using Convolutional Neural Networks.

Alzheimer's disease and mild cognitive impairment are common diseases in the elderly, affecting more than 50 million people worldwide in 2020. Early diagnosis is crucial for managing these diseases, but their complexity poses a challenge. Convolutional neural networks have shown promise in accurate diagnosis. The main objective of this research is to diagnose Alzheimer's disease and mild cognitive impairment in healthy individuals using convolutional neural networks. This study utilized three different convolutional neural network models, two of which were pre-trained models, namely AlexNet and DenseNet, while the third model was a CNN1D-LSTM neural network. Among the neural network models used, the AlexNet demonstrated the highest accuracy, exceeding 98%, in diagnosing mild cognitive impairment and Alzheimer's disease in healthy individuals. Furthermore, the accuracy of the DenseNet and CNN1D-LSTM models is 88% and 91.89%, respectively. The research highlights the potential of convolutional neural networks in diagnosing mild cognitive impairment and Alzheimer's disease. The use of pre-trained neural networks and the integration of various patient data contribute to achieving accurate results. The high accuracy achieved by the AlexNet neural network underscores its effectiveness in disease classification. These findings pave the way for future research and improvements in the field of diagnosing these diseases using convolutional neural networks, ultimately aiding in early detection and effective management of mild cognitive impairment and Alzheimer's disease.

Recognition of life-threatening arrhythmias by ECG scalograms

This work is devoted to the automatic classification of six classes of life-threatening arrhythmias using short ECG fragments of 2s-length. This task is extremely important for the detection of life-threatening arrhythmias with continuous monitoring. Especially dangerous are ventricular fibrillation and high-frequency heartbeat ventricular tachycardia. Timely detection of these dangerous disorders in the clinic allows doctors to effectively use electrical defibrillation, which saves the patient's life. A feature of our approach is the use of a unique technique for converting ECG signals into images (scalograms) using a continuous wavelet transform. For arrhythmia classification, the AlexNet neural network with a well-known deep learning architecture, which is commonly used in image classification tasks, is used. The experiments used data from the PhysioNet database, as well as synthesized ECG data obtained using the SMOTE method. The experimental results show that the proposed approach allows achieving an average accuracy of 98.7% for all classes, which exceeds the maximum accuracy estimates of 93.18% previously obtained by other researchers.

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.

Research on power equipment troubleshooting based on improved AlexNet neural network

Power equipment is an important component of the whole power system, so that it is obvious that it is required to develop a correct method for accurate analysis of the infrared image features of the equipment in the field of detection and recognition. This study proposes a troubleshooting strategy for the power equipment based on the improved AlexNet neural network. Multi-scale images based on the Pan model are used to determine the equipment features, and to determine the shortcomings of AlexNet neural network, such as slower recognition speed and easy overfitting. After knowing these shortcomings, it would become possible to improve the specific recognition model performance by adding a pooling layer, modifying the activation function, replacing the LRN with BN layer, and optimizing the parameters of the improved WOA algorithm, and other measures. In the simulation experiments, this paper's algorithm was compared with AlexNet, YOLO v3, and Faster R-CNN algorithms in the lightning arrester fault detection, circuit breaker fault detection, mutual transformer fault detection, and insulator fault detection improved by an average of 5.47 %, 4.69 %, and 3.42 %, which showed that the algorithm had a better recognition effect.

Automated facial characterization and image retrieval by convolutional neural networks.

Developing efficient methods to infer relations among different faces consisting of numerous expressions or on the same face at different times (e.g., disease progression) is an open issue in imaging related research. In this study, we present a novel method for facial feature extraction, characterization, and identification based on classical computer vision coupled with deep learning and, more specifically, convolutional neural networks. We describe the hybrid face characterization system named FRetrAIval (FRAI), which is a hybrid of the GoogleNet and the AlexNet Neural Network (NN) models. Images analyzed by the FRAI network are preprocessed by computer vision techniques such as the oriented gradient-based algorithm that can extract only the face region from any kind of picture. The Aligned Face dataset (AFD) was used to train and test the FRAI solution for extracting image features. The Labeled Faces in the Wild (LFW) holdout dataset has been used for external validation. Overall, in comparison to previous techniques, our methodology has shown much better results on k-Nearest Neighbors (KNN) by yielding the maximum precision, recall, F1, and F2 score values (92.00, 92.66, 92.33, and 92.52%, respectively) for AFD and (95.00% for each variable) for LFW dataset, which were used as training and testing datasets. The FRAI model may be potentially used in healthcare and criminology as well as many other applications where it is important to quickly identify face features such as fingerprint for a specific identification target.

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.

Flame Recognition and Localization Based on Convolutional Neural Network

Target detection and flame identification play a critical role in the industry. Unfortunately, current inspection systems are unable to reliably and easily identify flames. For this reason,, a convolutional neural network Alexnet is proposed for target detection. A webcam approach is used for image capture and collection, followed by network training. The target detection and recognition of flames are finally completed. Experimental results demonstrate that the algorithm as well as the Alexnet network can efficiently and accurately identify localized flames.

Use of covariance matrix images for electroencephalography signal classification for multiclass motor imagery‐based brain computer interface

AbstractElectroencephalogram (EEG) signals can be used to control devices using brain computer interface (BCI) systems, which offer assistance to patients with neuromuscular diseases. This article aims to determine the feasibility of using spatial covariance matrices with deep convolutional neural networks for multiclass motor imagery (MI) task classification for EEG‐based BCIs. Covariance matrices derived from multichannel EEG data are converted into color‐scaled images, which are used to train the AlexNet neural network using transfer learning. Samples of 2‐s duration are extracted every 200 ms intervals from each MI activity EEG eventually increasing the training data size sixfold and reducing the latency in real‐time implementations. Observed the performance using two popular publicly available BCI datasets for two‐class and four‐class MI classification. BCI competition IV Dataset IIa has shown average accuracy of 93.99% (k = 0.88) for two‐class and 86.70% (k = 0.82) for four‐class MI classification. BCI competition III Dataset IVa exhibited average accuracy of 97.02% (k = 0.94) for two‐class classification. Good performance was observed for training with cross‐subject data for two‐class and four‐class classification. The proposed method has shown improved accuracy for multiclass MI classification compared with existing works. The CNN is able to extract features from the covariance matrix and classify the MI data with reasonable accuracy. Since this method can handle large‐dimension data, dimension reduction, and source separation techniques are not required.

Apple Surface Defect Detection Method Based on Weight Comparison Transfer Learning with MobileNetV3

Apples are ranked third, after bananas and oranges, in global fruit production. Fresh apples are more likely to be appreciated by consumers during the marketing process. However, apples inevitably suffer mechanical damage during transport, which can affect their economic performance. Therefore, the timely detection of apples with surface defects can effectively reduce economic losses. In this paper, we propose an apple surface defect detection method based on weight contrast transfer and the MobileNetV3 model. By means of an acquisition device, a thermal, infrared, and visible apple surface defect dataset is constructed. In addition, a model training strategy for weight contrast transfer is proposed in this paper. The MobileNetV3 model with weight comparison transfer (Weight Compare-MobileNetV3, WC-MobileNetV3) showed a 16% improvement in accuracy, 14.68% improvement in precision, 14.4% improvement in recall, and 15.39% improvement in F1-score. WC-MobileNetV3 compared to MobileNetV3 with fine-tuning improved accuracy by 2.4%, precision by 2.67%, recall by 2.42% and F1-score by 2.56% compared to the classical neural networks AlexNet, ResNet50, DenseNet169, and EfficientNetV2. The experimental results show that the WC-MobileNetV3 model adequately balances accuracy and detection time and achieves better performance. In summary, the proposed method achieves high accuracy for apple surface defect detection and can meet the demand of online apple grading.

Evolution-by-Coevolution of Neural Networks for Audio Classification

Neural networks are increasingly used in recognition problems, including static and moving images, sounds, etc. Unfortunately, the selection of optimal neural network architecture for a specific recognition problem is a difficult task, which often has an experimental nature. In this paper we present the use of evolutionary algorithms to obtain optimal architectures of neural networks used for audio sample classification. We extend the Pytorch DNN Evolution tool implementing co-evolutionary algorithms which create groups of neural networks that solve a given problem with a certain accuracy, with the support for problems in which training data consists of audio samples. In this paper we use the co-evolutionary approach to solve a sample sound classification problem. We describe how the sound data was prepared for processing with the use of the Mel Frequency Cepstral Coefficients (MFCC). Next we present the results of experiments conducted with the AudioMnist dataset. The obtained neural network architectures, whose classification accuracy is comparable to the classification accuracy attained by the AlexNet neural network, and their implications are discussed.

HeLayers: A Tile Tensors Framework for Large Neural Networks on Encrypted Data

Privacy-preserving solutions enable companies to offload confidential data to third-party services while fulfilling their government regulations. To accomplish this, they leverage various cryptographic techniques such as Homomorphic Encryption (HE), which allows performing computation on encrypted data. Most HE schemes work in a SIMD fashion, and the data packing method can dramatically affect the running time and memory costs. Finding a packing method that leads to an optimal performant implementation is a hard task. We present a simple and intuitive framework that abstracts the packing decision for the user. We explain its underlying data structures and optimizer, and propose a novel algorithm for performing 2D convolution operations. We used this framework to implement an inference operation over an encrypted HE-friendly AlexNet neural network with large inputs, which runs in around five minutes, several orders of magnitude faster than other state-of-the-art non-interactive HE solutions.

Drought stress impact on the performance of deep convolutional neural networks for weed detection in Bahiagrass

AbstractMachine vision‐based weed detection relies on features such as plant colour, leaf texture, shape, and patterns. Drought stress in plants can alter leaf colour and morphological features, which may in turn affect the reliability of machine vision‐based weed detection. The objective of this research was to evaluate the feasibility of using deep convolutional neural networks for the detection of Florida pusley (Richardia scabra L.) growing in drought stressed and unstressed bahiagrass (Paspalum natatum Flugge). The object detection neural networks you only look once (YOLO)v3, faster region‐based convolutional network (Faster R‐CNN), and variable filter net (VFNet) failed to effectively detect Florida pusley growing in drought stressed or unstressed bahiagrass, with F1 scores ≤0.54 in the testing dataset. Nevertheless, the use of the image classification neural networks AlexNet, GoogLeNet, and Visual Geometry Group‐Network (VGGNet) was highly effective and achieved high (≥0.97) F1 scores and recall values (≥0.98) in detecting images containing Florida pusley growing in drought stressed or unstressed bahiagrass. Overall, these results demonstrated the effectiveness of using an image classification convolutional neural network for detecting Florida pusley in drought stressed or unstressed bahiagrass. These findings illustrate the broad applicability of these neural networks for weed detection.

Contactless Fall Detection by Means of Multiple Bioradars and Transfer Learning.

Fall detection in humans is critical in the prevention of life-threatening conditions. This is especially important for elderly people who are living alone. Therefore, automatic fall detection is one of the most relevant problems in geriatrics. Bioradiolocation-based methods have already shown their efficiency in contactless fall detection. However, there is still a wide range of areas to improve the precision of fall recognition based on view-independent concepts. In particular, in this paper, we propose an approach based on a more complex multi-channel system (three or four bioradars) in combination with the wavelet transform and transfer learning. In the experiments, we have used several radar configurations for recording different movement types. Then, for the binary classification task, a pre-trained convolutional neural network AlexNet has been fine-tuned using scalograms. The proposed systems have shown a noticeable improvement in the fall recognition performance in comparison with the previously used two-bioradar system. The accuracy and Cohen’s kappa of the two-bioradar system are 0.92 and 0.86 respectively, whereas the accuracy and Cohen’s kappa of the four-bioradar system are 0.99 and 0.99 respectively. The three-bioradar system’s performance turned out to be in between two of the aforementioned systems and its calculated accuracy and Cohen’s kappa are 0.98 and 0.97 respectively. These results may be potentially used in the design of a contactless multi-bioradar fall detection system.

A comparative evaluation of convolutional neural networks, training image sizes, and deep learning optimizers for weed detection in alfalfa

AbstractIn this research, the deep-learning optimizers Adagrad, AdaDelta, Adaptive Moment Estimation (Adam), and Stochastic Gradient Descent (SGD) were applied to the deep convolutional neural networks AlexNet, GoogLeNet, VGGNet, and ResNet that were trained to recognize weeds among alfalfa using photographic images taken at 200×200, 400×400, 600×600, and 800×800 pixels. An increase in the image sizes reduced the classification accuracy of all neural networks. The neural networks that were trained with images of 200×200 pixels resulted in better classification accuracy than the other image sizes investigated here. The optimizers AlexNet and GoogLeNet trained with AdaDelta and SGD outperformed the Adagrad and Adam optimizers; VGGNet trained with AdaDelta outperformed Adagrad, Adam, and SGD; and ResNet trained with AdaDelta and Adagrad outperformed the Adam and SGD optimizers. When the neural networks were trained with the best-performing input image size (200×200 pixels) and the best-performing deep learning optimizer, VGGNet was the most effective neural network, with high precision and recall values (≥0.99) when validation and testing datasets were used. Alternatively, ResNet was the least effective neural network in its ability to classify images containing weeds. However, there was no difference among the different neural networks in their ability to differentiate between broadleaf and grass weeds. The neural networks discussed herein may be used for scouting weed infestations in alfalfa and further integrated into the machine vision subsystem of smart sprayers for site-specific weed control.

FACIAL EXPRESSION RECOGNITION BASED ON FEATURE ENHANCEMENT AND IMPROVED ALEXNET

For interpersonal relations among humans, facial expressions are extremely important. Due to the complications in collecting required features from the frequently changing surroundings, uneven reflection from light sources, and many other aspects, facial expression recognition will encounter significant problems. A novel facial image recognition approach is proposed in this paper. Initially, a face image enhancement framework is created that is capable of enhancing the features of a face in a complicated context for this strategy. The improved Alexnet neural network is then created, which is based on the Alexnet architecture. Multi-scale convolution process is utilised in the improved Alexnet to enhance feature extraction capability. Batch normalisation is used for preventing network overfitting while also improving the model’s robustness. The Adabound optimizer and the Relu activation function are used to improve convergence and accuracy. The facial image feature improvement approach is helpful to increasing the capability of the improved Alexnet in trials from many aspects. For face images acquired in the natural surroundings, our technique displays significant stability, serving as a benchmark for the intelligent prediction of other facial images.

Approach to Automated Visual Inspection of Objects Based on Artificial Intelligence

The article discusses the possibility of object detector usage in field of automated visual inspection for objects with specific parameters, specifically various types of defects occurring on the surface of a car tire. Due to the insufficient amount of input data, as well as the need to speed up the development process, the Transfer Learning principle was applied in a designed system. In this approach, the already pre-trained convolutional neural network AlexNet was used, subsequently modified in its last three layers, and again trained on a smaller sample of our own data. The detector used in the designed camera inspection system with the above architecture allowed us to achieve the accuracy and versatility needed to detect elements (defects) whose shape, dimensions and location change with each occurrence. The design of a test facility with the application of a 12-megapixel monochrome camera over the rotational table is briefly described, whose task is to ensure optimal conditions during the scanning process. The evaluation of the proposed control system with the quantification of the recognition capabilities in the individual defects is described at the end of the study. The implementation and verification of such an approach together with the proposed methodology of the visual inspection process of car tires to obtain better classification results for six different defect classes can be considered as the main novel feature of the presented research. Subsequent testing of the designed system on a selected batch of sample images (containing all six types of possible defect) proved the functionality of the entire system while the highest values of successful defect detection certainty were achieved from 85.15% to 99.34%.

Advance Genome Disorder Prediction Model Empowered With Deep Learning

A major and essential issue in biomedical research is to predict genome disorder. Genome disorders cause multivariate diseases like cancer, dementia, diabetes, cystic fibrosis, leigh syndrome, etc. which are causes of high mortality rates around the world. In past, theoretical and explanatory-based approaches were introduced to predict genome disorder. With the development of technology, genetic data were improved to cover almost genome and protein then machine and deep learning-based approaches were introduced to predict genome disorder. Parallel machine and deep learning approaches were introduced. In past, many types of research were conducted on genome disorder prediction using supervised, unsupervised, and semi-supervised learning techniques, most of the approaches using binary problem prediction using genetic sequence data. The prediction results of these approaches were uncertain because of their lower accuracy rate and binary class prediction techniques using genome sequence data but not genome disorder patients&#x2019; data with his/her history. Most of the techniques used Ribonucleic acid (RNA) gene sequence and were not often capable of handling bid data effectively. Consequently, in this study, the AlexNet as an effective convolutional neural network architecture proposed to develop an advance genome disorder prediction model (AGDPM) for predicting genome multi classes disorder using a large amount of data. AGDPM tested and compare with the pre-trained AlexNet neural network model and AGDPM gives the best results with 89.89&#x0025; &#x0026; 81.25&#x0025; accuracy of training and testing respectively. So, the advance genome disorder prediction model shows the ability to efficiently predict genome disorder and can process a large amount of patients&#x2019; genome disorder data with a multi-class prediction method. AGDPM has proved that it is capable to predict single gene inheritance disorder, mitochondrial gene inheritance disorder, and multifactorial gene inheritance disorder with respect to various statistical performance parameters. So, with the help of AGDPM biomedical research will be improved in terms to predict genetic disorders and put control on high mortality rates.

To the question of realization of machine vision technology based on FPGA-architectures

In the last few years, machine learning and machine vision technologies have started to gain more and more popularity. This industry occupies one of the leading positions in the field of information technology. The paper is devoted to the development of a machine vision algorithm based on new generations of FPGAs for recognizing handwritten Cyrillic characters in images and video streams, in particular. The article raises the issues of using FPGA as an image segmentation accelerator and organizing work with the video stream, choosing the most suitable FPGA platform, creating training samples of handwritten characters, and working with the convolutional neural network AlexNet.