A New Reinforcement Learning Based Learning Rate Scheduler for Convolutional Neural Network in Fault Classification

Prediction of PM2.5 concentration in urban agglomeration of China by hybrid network model

The ensemble learning method is a necessary process that provides robustness and is more accurate than the single model. The snapshot ensemble convolutional neural network (CNN) has been successful and widely used in many domains, such as image classification, fault diagnosis, and plant image classification. The advantage of the snapshot ensemble CNN is that it combines the cyclic learning rate schedule in the algorithm to snap the best model in each cycle. In this research, we proposed the dropCyclic learning rate schedule, which is a step decay to decrease the learning rate value in every learning epoch. The dropCyclic can reduce the learning rate and find the new local minimum in the subsequent cycle. We evaluated the snapshot ensemble CNN method based on three learning rate schedules: cyclic cosine annealing, max-min cyclic cosine learning rate scheduler, and dropCyclic then using three backbone CNN architectures: MobileNetV2, VGG16, and VGG19. The snapshot ensemble CNN methods were tested on three aerial image datasets: UCM, AID, and EcoCropsAID. The proposed dropCyclic learning rate schedule outperformed the other learning rate schedules on the UCM dataset and obtained high accuracy on the AID and EcoCropsAID datasets. We also compared the proposed dropCyclic learning rate schedule with other existing methods. The results show that the dropCyclic method achieved higher classification accuracy compared with other existing methods.

Background: Memristor-based implementations of Convolutional Neural Networks (CNNs) offer the potential to reduce power consumption and improve computational efficiency significantly. However, the non-ideal characteristics of memristive devices, such as variability and limited switching behavior, can introduce nonlinearity and degrade network performance. It is therefore essential to simulate memristor-based deep learning systems and evaluate the impact of these imperfections before proceeding with hardware-level deployment. Objective: This study aims to design a lightweight memristor-based MobileNetV2 model for smoking behavior detection and analyze the influence of memristor non-idealities on the network’s detection accuracy. Methods: A lightweight MobileNetV2 network was constructed and trained on a self-developed dataset for smoking behavior detection, achieving a peak accuracy of 92.25%. Based on the trained model, a memristor-based MobileNetV2 architecture was implemented using the Mem- Torch simulation framework. Two memristor models, VTEAM and Data-Driven Verilog-A RRAM, were employed to replace the convolutional and fully connected layers with memristor crossbar arrays. The impact of non-ideal memristor properties, including conductance variability and resistance state faults, was systematically evaluated. Results: Simulation results demonstrated that the memristor-based MobileNetV2 network achieved detection accuracies of 92.11% with the VTEAM model and 91.82% with the Verilog-A RRAM model. As device non-idealities increased, the detection accuracy declined significantly, indicating a strong correlation between memristor imperfections and the performance of CNN- based target detection tasks. Conclusion: Memristor non-idealities substantially affect the performance of neuromorphic CNNs. This study provides targeted analyses and proposes mitigation strategies, such as differential encoding and improved fabrication techniques, to support the practical deployment of memristor- based neural networks in energy-efficient AI applications.

Every year, new Convolutional Neural Network (CNN) architectures appear to deal with different problems in the activity of image and video recognition. These architectures usually work along the ImageNet dataset for looking for the best performance of the CNNs without taking into account the video task where they are used. This can represent a problem if the task is Human Action Recognition (HAR) in video, since the CNN architectures are pre-trained with an image dataset that can practically contain any object, while HAR problem requires consecutive frames of people doing actions. To prove the idea that using CNNs pre-trained on an image dataset does not always achieve the best performance on a video dataset and that, therefore, it is worth comparing the performance of different CNNs under similar circumstances for the HAR problem, this work proposes an analysis between eight different CNN architectures. Each one of the CNN was exclusively trained with RGB images, which were extracted from the frames of the different classes of videos of HMDB51 dataset. To make the classification of an activity in video, we average the predictions taking into account the successes. We also made some ensembles with the best performance CNNs to measure the improvement in accuracy. Our results suggest that Xception is a strong baseline model that could be used by the community to make their comparisons of their proposals more robust.

Mild cognitive impairment (MCI) is a clinical condition characterized by a decline in cognitive ability and progression of cognitive impairment. It is often considered a transitional stage between normal aging and Alzheimer's disease (AD). This study aimed to compare deep learning (DL) and traditional machine learning (ML) methods in predicting MCI using plasma proteomic biomarkers. A total of 239 adults were selected from the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort along with a pool of 146 plasma proteomic biomarkers. We evaluated seven traditional ML models (support vector machines (SVMs), logistic regression (LR), naïve Bayes (NB), random forest (RF), k-nearest neighbor (KNN), gradient boosting machine (GBM), and extreme gradient boosting (XGBoost)) and six variations of a deep neural network (DNN) model-the DL model in the H2O package. Least Absolute Shrinkage and Selection Operator (LASSO) selected 35 proteomic biomarkers from the pool. Based on grid search, the DNN model with an activation function of "Rectifier With Dropout" with 2 layers and 32 of 35 selected proteomic biomarkers revealed the best model with the highest accuracy of 0.995 and an F1 Score of 0.996, while among seven traditional ML methods, XGBoost was the best with an accuracy of 0.986 and an F1 Score of 0.985. Several biomarkers were correlated with the APOE-ε4 genotype, polygenic hazard score (PHS), and three clinical cerebrospinal fluid biomarkers (Aβ42, tTau, and pTau). Bioinformatics analysis using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) revealed several molecular functions and pathways associated with the selected biomarkers, including cytokine-cytokine receptor interaction, cholesterol metabolism, and regulation of lipid localization. The results showed that the DL model may represent a promising tool in the prediction of MCI. These plasma proteomic biomarkers may help with early diagnosis, prognostic risk stratification, and early treatment interventions for individuals at risk for MCI.

Mobile fitness applications provide the opportunity to show users real-time feedback on their current fitness activity. For such applications, it is essential to accurately track the user's current fitness activity using available mobile sensors, such as inertial measurement units (IMUs). Convolutional neural networks (CNNs) have been shown to produce strong results in different time series classification tasks, including the recognition of daily living activities. However, fitness activities can present unique challenges to the human activity recognition task (HAR), including greater similarity between individual activities and fewer available data for model training. In this paper, we evaluate the applicability of CNNs to the fitness activity recognition task (FAR) using IMU data and determine the impact of input data size and sensor count on performance. For this purpose, we adapted three existing CNN architectures to the FAR task and designed a fourth CNN variant, which we call the scaling fully convolutional network (Scaling-FCN). We designed a preprocessing pipeline and recorded a running exercise data set with 20 participants, in which we evaluated the respective recognition performances of the four networks, comparing them with three traditional machine learning (ML) methods commonly used in HAR. Although CNN architectures achieve at least 94% test accuracy in all scenarios, two traditional ML architectures surpass them in the default scenario, with support vector machines (SVMs) achieving 99.00 ± 0.34% test accuracy. The removal of all sensors except one foot sensor reduced the performance of traditional ML architectures but improved the performance of CNN architectures on our data set, with our Scaling-FCN reaching the highest accuracy of 99.86 ± 0.11% on the test set. Our results suggest that CNNs are generally well suited for fitness activity recognition, and noticeable performance improvements can be achieved if sensors are dropped selectively, although traditional ML architectures can still compete with or even surpass CNNs when favorable input data are utilized.

Deep learning theory demonstrates its great power in the complicated data processing field. As a prevalent deep learning technology, convolutional neural network (CNN) has achieved some successful applications in the field of fault detection and classification in the complicated industrial systems because of its strong nonlinear feature extraction capability. However, the basic CNN views all the features equally and can not highlight these features which play a more important role in the fault classification procedure. To deal with this problem, this paper proposes an improved CNN model, called multi-head attention CNN (MACNN), to distinguish the importance of different features for better fault classification performance. In the MACNN framework, the raw training data are firstly analyzed by the multiple convolutional layers for feature extraction. Then, the features obtained by the convolution layers are used as the input of the multi-head attention (MA) layers. The introduction of the MA mechanism can help exploit more meaningful information by emphasizing the important features. Lastly, the softmax classifier is utilized to classify the fault patterns. To test the method performance, one case study on the benchmark Tennessee Eastman industrial system is performed and the results show that MACNN gives the better fault classification results than the traditional feedforward neural network (FNN) and CNN models.

Diabetic retinopathy (DR) is one of the leading causes of vision loss in adults and is one of the detrimental side effects of the mass prevalence of Diabetes Mellitus (DM). It is crucial to have an efficient screening method for early diagnosis of DR to prevent vision loss. This paper compares and analyzes the various Machine Learning (ML) techniques, from traditional ML to advanced Deep Learning models. We compared and analyzed the efficacy of Convolutional Neural Networks (CNNs), Capsule Networks (CapsNet), K-Nearest Neighbor (KNN), Support Vector Machine (SVM), decision trees, and Random Forests. This paper also considers determining factors in the evaluation, including contrast enhancements, noise reduction, grayscaling, etc We analyze recent research studies and compare methodologies and metrics, including accuracy, precision, sensitivity, and specificity. The findings highlight the advanced performance of Deep Learning (DL) models, with CapsNet achieving a remarkable accuracy of up to 97.98% and a high precision rate, outperforming other traditional ML methods. The Contrast Limited Adaptive Histogram Equalization (CLAHE) preprocessing technique substantially enhanced the model’s efficiency. Each ML method’s computational requirements are also considered. While most advanced deep learning methods performed better according to the metrics, they are more computationally complex, requiring more resources and data input. We also discussed how datasets like MESSIDOR could be more straightforward and contribute to highly evaluated performance and that there is a lack of consistency regarding benchmark datasets across papers in the field. Using the DL models facilitates accurate early detection for DR screening, can potentially reduce vision loss risks, and improves accessibility and cost-efficiency of eye screening. Further research is recommended to extend our findings by building models with public datasets, experimenting with ensembles of DL and traditional ML models, and considering testing high-performing models like CapsNet.

Like many optimization algorithms, stochastic variational inference is sensitive to the choice of the learning rate. If the learning rate is too small, the optimization process may be slow, and the algorithm might get stuck in local optima. On the other hand, if the learning rate is too large, the algorithm may oscillate or diverge, failing to converge to a solution. Adaptive learning rate methods such as Adam, AdaMax, Adagrad, or root mean square propagation automatically adjust the learning rate based on the history of gradients. Nevertheless, if the base learning rate is too large, the variational parameters might still oscillate around the optimal solution. With learning rate schedules, the learning rate can be reduced gradually to mitigate this problem. However, the amount at which the learning rate should be decreased in each iteration is not known a priori , which can significantly impact the performance of the optimization. In this work, we propose a method to decay the learning rate based on the history of the variational parameters. We use an empirical measure to quantify the amount of oscillations against the progress of the variational parameters to adapt the learning rate. The approach requires little memory and is computationally efficient. We demonstrate in various numerical examples that our method reduces the sensitivity of the optimization performance to the learning rate and that it can also be used in combination with other adaptive learning rate methods.

Accurate identification and classification of equipment defects are essential for assessing the health of power equipment and making informed maintenance decisions. Traditional defect classification methods, which rely on subjective manual records and intricate defect descriptions, have proven to be inefficient. Existing approaches often evaluate equipment status solely based on defect grade. To enhance the precision of power equipment defect identification, we developed a multi-label classification dataset by compiling historical defect records. Furthermore, we assessed the performance of 11 established multi-label classification methods on this dataset, encompassing both traditional machine learning and deep learning methods. Experimental results reveal that methods considering label correlations exhibit significant performance advantages. By employing balanced loss functions, we effectively address the challenge of sample imbalance across various categories, thereby enhancing classification accuracy. Additionally, segmenting the power equipment defect classification task, which involves numerous labels, into label recall and ranking stages can substantially improve classification performance. The dataset we have created is available for further research by other scholars in the field.

Aims: This study aimed to explore the effect of training set diversity on the performance of deep learning models for predicting early gastric cancer (EGC) under endoscopy. Methods: Images of EGC and non-cancerous lesions under narrow-band imaging (ME-NBI) and magnifying blue laser imaging (ME-BLI) were retrospectively collected. Training set 1 was composed of 150 non-cancerous and 309 EGC ME-NBI images, training set 2 was composed of 1505 non-cancerous and 309 EGC ME-BLI images, and training set 3 was the combination of training set 1 and 2. Test set 1 was composed of 376 non-cancerous and 1052 EGC ME-NBI images, test set 2 consisted of 529 non-cancerous and 71 EGC ME-BLI images, and test set 3 was the combination of test set 1 and test set 2. Three deep learning models, convolutional neural network (CNN) 1, CNN 2 and CNN 3 (CNN 1, CNN 2 and CNN 3 were independently trained using training set 1, training set 2 and training set 3, respectively), were constructed, and their performances on each test set were respectively evaluated. One hundred and thirty-eight ME-NBI videos and 17 ME-BLI videos were further collected to evaluate and compare the performance of each model in real time. Results: On the whole, the performance of CNN 3 was the best. The accuracy (Acc), sensitivity (Sn), specificity (Sp) and area under the curve (AUC) of test set 1 in CNN 3 were 87.89% (1255/1428), 90.96% (342/376), 86.79% (913/1052) and 94.60%, respectively. The Acc, Sn, Sp and AUC of test set 2 in CNN 3 were 95% (570/600), 97.92% (518/529), 73.24% (52/71) and 90.93% respectively. The Acc, Sn, Sp and AUC of test set 3 in CNN 3 were 89.99% (1825/2028), 95.03% (860/905), 85.93% (965/1123) and 94.89%, respectively. The performance of CNN 3 was also the best in videos test set. The Acc, Sn and Sp of videos test set in CNN 3 were 91.03% (142/156), 90.58% (125/138) and 94.44% (17/18), respectively. Conclusions: The deep learning model with the most diverse training data has the best diagnostic effect.

ABSTRACTDue to the ever-expanding volume of observed spectroscopic data from surveys such as SDSS and LAMOST, it has become important to apply artificial intelligence (AI) techniques for analysing stellar spectra to solve spectral classification and regression problems like the determination of stellar atmospheric parameters Teff, $\rm {\log g}$, and [Fe/H]. We propose an automated approach for the classification of stellar spectra in the optical region using convolutional neural networks (CNNs). Traditional machine learning (ML) methods with ‘shallow’ architecture (usually up to two hidden layers) have been trained for these purposes in the past. However, deep learning methods with a larger number of hidden layers allow the use of finer details in the spectrum which results in improved accuracy and better generalization. Studying finer spectral signatures also enables us to determine accurate differential stellar parameters and find rare objects. We examine various machine and deep learning algorithms like artificial neural networks, Random Forest, and CNN to classify stellar spectra using the Jacoby Atlas, ELODIE, and MILES spectral libraries as training samples. We test the performance of the trained networks on the Indo-U.S. Library of Coudé Feed Stellar Spectra (CFLIB). We show that using CNNs, we are able to lower the error up to 1.23 spectral subclasses as compared to that of two subclasses achieved in the past studies with ML approach. We further apply the trained model to classify stellar spectra retrieved from the SDSS data base with SNR > 20.

The research aims to develop an advanced and precise lung cancer screening model based on Convolutional Neural Networks (CNN). Based on the health medical big data platform of Shandong University, we developed a VGG16-Based CNN lung cancer screening model. This model was trained using the Computed Tomography scans data of patients from Pingyi Traditional Chinese Medicine Hospital in Shandong Province, from January to February 2023. Data augmentation techniques, including random resizing, cropping, horizontal flipping, color jitter, random rotation and normalization, were applied to improve model generalization. We used five-fold cross-validation to robustly assess performance. The model was fine-tuned with an SGD optimizer (learning rate 0.001, momentum 0.9, and L2 regularization) and a learning rate scheduler. Dropout layers were added to prevent the model from relying too heavily on specific neurons, enhancing its ability to generalize. Early stopping was implemented when validation loss did not decrease over 10 epochs. In addition, we evaluated the model's performance with Area Under the Curve (AUC), Classification accuracy, Positive Predictive Value (PPV), and Negative Predictive Value (NPV), Sensitivity, Specificity and F1 score. External validation used an independent dataset from the same hospital, covering January to February 2022. The training and validation loss and accuracy over iterations show that both accuracy metrics peak at over 0.9 by iteration 15, prompting early stopping to prevent overfitting. Based on five-fold cross-validation, the ROC curves for the VGG16-Based CNN model, demonstrate an AUC of 0.963 ± 0.004, highlighting its excellent diagnostic capability. Confusion matrices provide average metrics with a classification accuracy of 0.917 ± 0.004, PPV of 0.868 ± 0.015, NPV of 0.931 ± 0.003, Sensitivity of 0.776 ± 0.01, Specificity of 0.962 ± 0.005 and F1 score of 0.819 ± 0.008, respectively. External validation confirmed the model's robustness across different patient populations and imaging conditions. The VGG16-Based CNN lung screening model constructed in this study can effectively identify lung tumors, demonstrating reliability and effectiveness in real-world medical settings, and providing strong theoretical and empirical support for its use in lung cancer screening.

The accurate estimation of aboveground biomass (AGB) in rubber plantations is essential for predicting rubber production and assessing carbon storage. Multispectral sensors mounted on unmanned aerial vehicles (UAVs) can obtain high spatiotemporal resolution imagery of rubber plantations, offering significant advantages in capturing fine structural details and heterogeneity. However, most previous studies primarily focused on developing biomass estimation models for rubber using machine learning (ML) algorithms in conjunction with feature selection methods based on UAV-acquired multispectral imagery. The reliance on feature selection methods limits the model’s generalizability, robustness, and predictive accuracy. In contrast, deep learning (DL) exhibits considerable promise in extracting features from high-resolution UAV-based multispectral imagery without the need for manual selection. Nonetheless, it remains unclear whether DL can surpass traditional ML methods in improving the AGB estimation accuracy in rubber plantations. To address this, our study evaluated the performance of three ML algorithms (random forest regression, RFR; XGBoost regression, XGBR; categorical boosting regression, CatBoost) combined with feature selection techniques and a deep convolutional neural network (DCNN) using multispectral imagery obtained from UAV for the AGB estimation of rubber plantations. The results indicate that the RFR combined with a principal component analysis (PCA) for feature selection yielded the best performance (R2 = 0.81, RMSE = 11.63 t/ha, MAE = 9.27 t/ha) between the three ML algorithms. Meanwhile, the DCNN model derived from the G, R, and NIR spectral bands achieved the highest estimation accuracy (R2 = 0.89, RMSE = 6.44 t/ha, MAE = 5.72 t/ha), where it outperformed the other ML methods. Our study highlights the great potential of combining UAV-based multispectral imagery with DL techniques to improve AGB estimation in rubber plantations, offering a new perspective for estimating the physiological and biochemical growth parameters of forests.

Ultrasound (US) is the preferred modality for fatty liver disease diagnosis due to its noninvasive, real-time, and cost-effective imaging capabilities. However, traditional B-mode US is qualitative, and therefore, the assessment is very subjective. Computer-aided diagnostic tools can improve the specificity and sensitivity of US and help clinicians to perform uniform diagnoses. In this work, we propose a novel deep learning model for nonalcoholic fatty liver disease classification from US data. We design a multi-feature guided multi-scale residual convolutional neural network (CNN) architecture to capture features of different receptive fields. B-mode US images are combined with their corresponding local phase filtered images and radial symmetry transformed images as multi-feature inputs for the network. Various fusion strategies are studied to improve prediction accuracy. We evaluate the designed network architectures on B-mode in vivo liver US images collected from 55 subjects. We also provide quantitative results by comparing our proposed multi-feature CNN architecture against traditional CNN designs and machine learning methods. Quantitative results show an average classification accuracy above 90% over tenfold cross-validation. Our proposed method achieves a 97.8% area under the ROC curve (AUC) for the patient-specific leave-one-out cross-validation (LOOCV) evaluation. Comprehensive validation results further demonstrate that our proposed approaches achieve significant improvements compared to training mono-feature CNN architectures ([Formula: see text]). Feature combination is valuable for the traditional classification methods, and the use of multi-scale CNN can improve liver classification accuracy. Based on the promising performance, the proposed method has the potential in practical applications to help radiologists diagnose nonalcoholic fatty liver disease.