Bearing Fault Diagnosis Method Based on Improved Convolutional Neural Network

Convolutional neural network is the most important algorithm in the field of deep learning. The traditional convolution neural network usually uses Sigmoid or Relu as the activation function, but the two sides of Sigmoid are saturated, and Relu has a dead zone, which is very easy to cause gradient disappearance and gradient explosion. In this paper, Mish activation function is introduced into LENET-5 convolutional neural network, which overcomes the shortcomings of traditional activation function. At the same time, the fixed memory step gradient descent method is used to replace the gradient descent method of the optimization part, which improves the global convergence of the algorithm.

Robust convolution kernel quantity determination based on corner radiation area adaptation

Many classical methods have been used in automatic sleep stage classification but few methods explore deep learning. Meanwhile, most deep learning methods require extensive expertise and suffer from a mass of handcrafted steps which are time-consuming. In this paper, we propose an efficient convolutional neural network, Sle-CNN, for five-sleep-stage classification. We attach each kernel in the first layers with a trainable coefficient to enhance the learning ability and flexibility of the kernel. Then, we make full use of the genetic algorithm’s heuristic search and the advantage of no need for the gradient to search for the sleep stage classification architecture. We verify the convergence of Sle-CNN and compare the performance of traditional convolutional neural networks before and after using the trainable coefficient. Meanwhile, we compare the performance between the Sle-CNN generated through genetic algorithm and the traditional convolutional neural networks. The experiments demonstrate that the convergence of Sle-CNN is faster than the normal convolutional neural networks and the Sle-CNN generated by genetic algorithm outperforms the traditional handcrafted counterparts too. Our research suggests that deep learning has a great potential on electroencephalogram signal processing, especially with the intensification of neural architecture search. Meanwhile, neural architecture search can exert greater power in practical engineering applications. We conduct the Sle-CNN with the Python library, Pytorch, and the code and models will be publicly available.

A new multi-data-driven spatiotemporal PM2.5 forecasting model based on an ensemble graph reinforcement learning convolutional network

The traditional convolutional neural network (CNN) only focuses on the features of the last layer of the network, ignoring those of other layers. The detailed information of the shallow network can improve the accuracy of classification to some extent. With the deepening of the network, the problems such as gradient disappearance and explosion are obvious. In order to alleviate this problem, a method for residual network image classification with multi-scale feature fusion is proposed in the paper. First of all, the Random Image Cropping and Patching (RICAP), a data augmentation method, is adopted to cut and splice the new training samples from the training set, based on which feature maps of different sizes are obtained by residual modules. Secondly, after reducing the dimension of high-level feature maps, all the feature maps with the same dimension are processed by the multi-scale fusion. Finally, these feature vectors are input into the Softmax classifier for training and classification, and the method of learning rate decay is used in the training process. The experiment results indicate that this method has a good performance on image classification based on the public datasets of MNIST and CIFAR-10.KeywordsMulti-scale feature fusionResidual networkData augmentationImage classification

<p class="Normal1">Medicinal plants have been found and utilized in traditional medical practices from ancient times. Many medicinal plants play a vital role in curating many life threatening diseases. Very few of the medicinal herbs are commercially cultivated. Many plant diseases are there which destroys these medicinal plants. Early detection of plant diseases can prevent the huge loss of these medicinal plants. Here, we presented a hybrid model that makes use of SVM along with the traditional convolutional neural network (CNN) for predicting <em>Basil</em> plants leaves diseases. We transformed the conventional CNN model by adding a classification layer Support Vector Machine (SVM) and K-Nearest Neighbor (KNN) after feature extraction and this approach tends to perform better than traditional CNN as we make the dataset balanced by data augmentation and SVN and KNN tend to perform better in case of balanced samples. CNN is used for training, SVM/KNN is used for classification. The advantages of CNN and SVM are used in proposed the CNN and SVM and KNN model. It is assumed that such a combined model would incorporate the benefits of CNN and SVM. Here, we identified the four types of diseases that affect <em>basil</em> plant leaves as Leaf spot, Downy mildew, Fusarium wilt, Fungal, and Healthy. Since there isn’t a standard dataset for <em>basil</em> leaves, we created our own 803 picture data set and used various machine learning techniques to train and evaluate the model. However, over other existing algorithms, our hybrid model i.e., CNN+SVM has produced more accurate results. For five classes of <em>basil</em> plant leaves, the proposed model produced 95.02% accuracy of for leaf diseases.</p>

Rolling bearing is an essential part of various rotating machines, and its signal is the typical nonlinear signal. Traditional fault diagnosis usually relies on manual experience to extract the features of signals first. Deep convolutional neural networks (DCNN) can make fuller use of time series than traditional convolutional neural networks (CNN). Because of the low accuracy rate, fault diagnosis using gated recurrent unit (GRU) alone is not unsatisfactory. In order to improve the temporality of one-dimensional convolutional neural networks (1D-CNN) and enhance the accuracy of GRU, a novel fault diagnosis method called deep convolutional neural networks - gated recurrent unit (DCNN-GRU) is first put forward, which combines DCNN with GRU. The original signals without preprocessing are input into the DCNN, and the outputs of DCNN are input into the GRU consequently. Then the faults of rolling bearing can be diagnosed effectively. As the post-processing method, the t-distributed stochastic neighbor embedding (t-SNE) method is applied to visualize the fault diagnosis results. Six different network models, including the DCNN-GRU, are used to train the same fault dataset for comparison. The simulation results show that the proposed method can reach more than 99.9% accuracy stably for the given dataset, which can verify the feasibility and effectiveness of proposed method. And the DCNN-GRU can also be verified with good generalization ability using different dataset.

Hyperspectral image classification based on spatial and spectral kernels generation network

Transformer-based models are driving a significant revolution in the field of machine learning at the moment. Among these innovations, vision transformers (ViTs) stand out for their application of transformer architectures to vision-related tasks. By demonstrating performance as good, if not better, than traditional convolutional neural networks (CNNs), ViTs have managed to capture considerable interest in the field. This study focuses on the resilience of ViTs and CNNs in the face of adversarial attacks. Such attacks, which introduce noise into the input of machine learning models to produce incorrect outputs, pose significant challenges to the reliability of machine learning models. Our analysis evaluated the adversarial robustness of CNNs and ViTs by using regularization techniques and adversarial training methods. Adversarial training, in particular, represents a traditional approach to boosting defenses against these attacks. Despite its prominent use, our findings reveal that regularization techniques enable vision transformers and, in most cases, CNNs to enhance adversarial defenses more effectively. Through testing datasets like CIFAR-10 and CIFAR-100, we demonstrate that vision transformers, especially when combined with effective regularization strategies, demonstrate adversarial robustness, even without adversarial training. Two main inferences can be drawn from our findings. Firstly, it emphasizes how effectively vision transformers could strengthen artificial intelligence defenses against adversarial attacks. Secondly, it shows how regularization, which requires much fewer computational resources and covers a wide range of adversarial attacks, can be effective for adversarial defenses. Understanding and improving a model’s resilience to adversarial attacks is crucial for developing secure, dependable systems that can handle the complexity of real-world applications as artificial intelligence and machine learning technologies advance.

In recent years, AI(Artificial Intelligence) has achieved great development in modern society. More and more modern technologies are applied in surveillance and monitoring. Healthcare monitoring is growing ubiquitous in modern wearable devices, such as a smart watch, electrocardiogram (ECG) necklace, smart band. Many sensors are attached to these smart devices to record and monitor physiological signals caused by activities, and then propagated those recorded electrical data to be further processed to give health diagnosis, disease prevention or making a distress call automatically. Obstructive sleep apnea (OSA) is a sleep disorder with a high occurrence in adult people and observed as an autonomous risk factor for circulatory problems such as ischemic heart attacks and stroke. Numerous traditional neural network based methods have been developed to detect OSA, where these methods however could not provide the intended result because they rely on shallow network. In this paper, we propose an effective OSA detection based on Convolutional neural network. Our method first extracts features from Apnea-Electrocardiogram (ECG) recordings using RR-intervals (time interval from one R-wave to the next R-wave in an ECG signal) and then CNN model having three convolution layers and three fully connected layers is trained with extracted features and applied for OSA detection. The first two convolution layers are followed by batch normalization and pooling layer, and softmax is connected to the last fully connected layer to give final decision. Experimental results on extracted feature of Apnea-ECG signal reveal that our model have better results in terms of performance measure sensitivity, specificity and accuracy. It is expected that the related technology can be applied into smart sensors, especially wearable devices.

In order to realize the position detection of false data injection, a position detection model based on bi-directional gated cycle unit optimized full convolution neural network is proposed. In view of the problem that the traditional false data injection detection is limited to judging whether the attack occurs or not, the processing of multi-level detection is proposed to realize the fine discrimination of the attacked data and effectively avoid the resource loss caused by direct location detection. According to the time series characteristics of the measured data, the bi-directional gated cycle unit is applied to the convolution layers of the full convolution neural network, which can effectively extract the spatiotemporal features between the data and improve the accuracy and efficiency of false data injection detection. The experimental results show that the optimized position detection model has a certain improvement in detection accuracy and efficiency compared with the traditional convolution neural network.

Traditional convolutional neural networks (CNNs) typically rely on a fully connected or dense layer for final decision-making, where each neuron connects to all neurons in the previous layer. The softmax function converts these outputs into a probability distribution over the target classes. While effective, this approach may limit the network ability to capture complex image details across multiple levels of abstraction. Shallow architectures may suffice for simple problems, whereas deeper networks are often required for more complicated tasks. A straightforward solution to improve classification is constructing multiple CNN branch models and using a naïve ensemble method that combines their outputs equally. However, this approach can underperform when weaker models improperly influence the final prediction. Unlike greedy ensemble growing that exhaustively evaluates all candidate models to identify the optimal subset, in this paper we propose an iterative ensemble threshold selection technique that progressively selects optimal thresholds to integrate only high-confidence predictions. By leveraging diverse CNN structures, our method captures features at varying complexities, improves classification robustness, and reduces computational overhead. Experimental results on two benchmark datasets demonstrate that our approach matches the accuracy of greedy ensemble methods while requiring significantly fewer evaluations, outperforming conventional CNN architectures.

According to the problem of overfitting in the traditional Convolutional Neural Networks(CNN) with large samples, an improved sparse DropConnect algorithm is put forward. With proposed algorithm, the CNN is improved and optimized. The performance of the network in image recognition is verified. Experiments of weld image recognition and classification indicate that the selective ability of the CNN on the sparse feature is improved. High recognition accuracy and classification accuracy of test sample are achieved. Over-fitting in the traditional network model can be avoided. More important, the feature sparsity and discriminative ability of the CNN are increased.

Objectives: The paper presents application of convolution neural network and artificial neural network for image classification problem for clothing dataset along with their performance comparison against different optimizers. The major objective of this paper is to perform image classification on fashion-mnist clothing dataset images. Methods: The methods used here are, the traditional ANN and CNN. Here image classification is performed on Fashion-mnist, clothing dataset using CNN and ANN with different optimizers. The performance of the working of ANN and CNN in classifying images from fashion-mnist dataset is compared against different optimizers namely stochastic gradient Descent, Adagrad, RMS prop and Adam optimizer. Findings: The study found that CNN worked better than ANN yielding training accuracy of 95%, 93% and testing accuracy of 91%, 89% when used with Adam and RmsProp respectively. Novelty: The novelty of this work is to present a comparative study of image classification using CNN, ANN using different optimizers, since not many studies or research articles showed the performance comparison of traditional and convolution neural networks in image classification along with different optimizers. Since the real-world scenarios of today require enormous data to be processed, CNN can fit well to diversify applications since they highly reduce the number of parameters to be trained that speeds up the training process. Moreover, to be specific on image classification problems they require the best and most prominent features to be detected and uncovered; this can be achieved using CNN since it has the concept of convolution using filters at its Core. Hence, CNN is highly recommended for such image classification applications than the traditional artificial-neural-networks because of the aforementioned reasons. Keywords: ANN (Artificial NeuralNetworks); CNN (ConvolutionNeural Networks); Optimization; SGD (Stochastic Gradient Descent); RmsProp (Root mean Square propagation); Adam (Adaptive moment estimation); AdaGrad (Adaptive Gradient)

Around 30% revenue has been wasted due to different types of advertising fraud. In view of the continuous emergence of fraud and the improvement of fraudulent means, a click fraud detection of cost-sensitive convolutional neural network (CSCNN) based on feature matrix is proposed. In order to capture the pattern of click fraud, priori probability is introduced and a new feature set is derived from the knowledge of entropy. Then, using cost-based sampling method and threshold-moving cost sensitive learning algorithm to solve unbalanced data training problems. Features are transformed into feature matrix by different time windows to fit the input of a convolutional neural network. Finally, considering the convolution kernel size, the number of convolution kernels and the pooling size of different network structures in this work, effective CSCNN structure was chosen. Experiments on real-word data show that this method can complete click fraud detection effectively, and can complete multi classification without increasing the complexity of the model. A new effective method was provided on click fraud detection in mobile advertisement.