3D-CNN Model Research Articles

Semi-quantitative identification of pore characteristics for Yungang grotto sandstone via acoustic emission signals and information fusion convolutional neural network

Diseases of grotto can be identified via artificial intelligence (AI) techniques, but most existing AI models cannot quantify the specific Indicators, such as pore parameters. In this work, an information fusion convolutional neural network (IFCNN) integrating acoustic emission (AE) time and frequency domain signals is constructed based on quantifiable labels of pore characteristics. And the model training process is introduced and the identification performance of IFCNN was compared with that of other machine learning methods. The specific pore indicators and its corresponding refined parameter range can inverted via six weathering levels in quantifiable labels. The IFCNN model is trained by using 6048 time-frequency domain AE series data and IFCNN demonstrated superior identification performance and generalization ability compared to other 1D-CNN and machine learning models. Furthermore, 3 extra sandstone samples with different degrees of weathering are used to further investigate the pore parameter recognition performance of the IFCNN pre-trained model. It's found that the effective T2 intervals of the 3 samples were all identified incorrectly, that is, micro fissures intervals greater than 200 ms were missed. However, the accuracy of identified total bulk porosity, average pore size, and pore development coefficient is better. Except for average pore size, all other indicators are within the given interval from semi-quantified label. When the pore size distribution is about 0.5 μm–10 μm, the weathering of Yungang sandstone can be described semi-quantitatively. It is envisioned that the proposed IFCNN serves as an accurate and ready-to-use assessment tool that aids researchers or cultural heritage conservators to quickly evaluate the pore parameters of sandstone grottoes just using AE device.

Chlorophyll Content Estimation of Ginkgo Seedlings Based on Deep Learning and Hyperspectral Imagery

Accurate estimation of chlorophyll content is essential for understanding the growth status and optimizing the cultivation practices of Ginkgo, a dominant multi-functional tree species in China. Traditional methods based on chemical analysis for determining chlorophyll content are labor-intensive and time-consuming, making them unsuitable for large-scale dynamic monitoring and high-throughput phenotyping. To accurately quantify chlorophyll content in Ginkgo seedlings under different nitrogen levels, this study employed a hyperspectral imaging camera to capture canopy hyperspectral images of seedlings throughout their annual growth periods. Reflectance derived from pure leaf pixels of Ginkgo seedlings was extracted to construct a set of spectral parameters, including original reflectance, logarithmic reflectance, and first derivative reflectance, along with spectral index combinations. A one-dimensional convolutional neural network (1D-CNN) model was then developed to estimate chlorophyll content, and its performance was compared with four common machine learning methods, including Gaussian Process Regression (GPR), Partial Least Squares Regression (PLSR), Support Vector Regression (SVR), and Random Forest (RF). The results demonstrated that the 1D-CNN model outperformed others with the first derivative spectra, achieving higher CV-R2 and lower RMSE values (CV-R2 = 0.80, RMSE = 3.4). Furthermore, incorporating spectral index combinations enhanced the model’s performance, with the 1D-CNN model achieving the best performance (CV-R2 = 0.82, RMSE = 3.3). These findings highlight the potential of the 1D-CNN model in strengthening the chlorophyll estimations, providing strong technical support for the precise cultivation and the fertilization management of Ginkgo seedlings.

A small-sized fire detection method based on the combination of the SIC algorithm and 1-DCNN

The article proposes a method for detecting small-sized fires using an Ultra-Weak Fiber Bragg Grating (UWFBG) array fire monitoring system, addressing issues of slow response speed and insufficient spatial resolution in traditional fire monitoring systems. The system’s spatial resolution is improved by regional subsection monitoring and using the Spectral Intensity Change (SIC) algorithm combined with a one-dimensional convolutional neural network (1-DCNN) model for hot spot recognition. This approach can ignore the impact of noise on the reflected spectra to some extent, overcoming the limitations of traditional methods. Simulation results demonstrate that when the signal-to-noise ratio (SNR) is above 10 dB, the accuracy of hot spot recognition is significantly improved. At an SNR of 15 dB, the recognition accuracy reaches 98.5 %. Furthermore, addressing the spectral distortion issue in the sensor array, introducing the 1-DCNN model achieves a recognition accuracy of 98.6 %. The proposed method can identify small-sized fires of 10 cm, providing a safe and reliable solution for fire monitoring systems in subway tunnels.

Multimodal fish maw type recognition based on Wasserstein generative adversarial network combined with gradient penalty and spectral fusion

There are many types of fish maw with significantly varying prices. The specific type directly affects its market value and medicinal efficacy. This paper proposes a fish maw type recognition method based on Wasserstein generative adversarial network combined with gradient penalty (WGAN-GP) and spectral fusion. By collecting Raman and near-infrared (NIR) spectral data of four types of fish maw (Beihai Male Fish Maw, Beihai Female Fish Maw, Yellow Croaker Fish Maw, and Red Mouth Croaker Fish Maw), we used WGAN-GP for data augmentation. The performance of three spectral fusion strategies (data layer, feature layer, and decision layer) was explored based on two one-dimensional convolutional neural network (1D-CNN) models. The results indicate that, after applying data augmentation and expanding the training set to 3,600 samples, the performances of the 1D-VGG (NIR), 1D-VGG (Raman), 1D-ResNet (NIR), and 1D-ResNet (Raman) models all reach optimal levels. The accuracies on the test set are improved by 15.48%, 13.10%, 1.19%, and 5.95%, respectively. Under different fusion strategies, the 1D-VGG (Raman)-1D-VGG (NIR) model at the feature layer and 1D-ResNet (Raman)(1.0)-1D-ResNet (NIR)(1.0) model at the decision layer achieved the same classification results. They exceeded other models in accuracy (98.21%), precision (98.27%), recall (98.21%), and F1-score (98.21%) on the test set. In summary, this study demonstrated the great potential of data enhancement and multimodal spectral data fusion in fish maw type identification, providing analytical tools for the development of fish maw detection equipment based on multimodal techniques.

Auxiliary diagnosis of primary bone tumors based on Machine learning model

ObjectiveResearch on auxiliary diagnosis of primary bone tumors can enhance diagnostic accuracy, facilitate early detection, and enable personalized treatment, thereby reducing misdiagnosis and missed cases, ultimately leading to improved patient prognosis and survival rates. In this study, we established a whole slide imaging (WSI) database comprising histopathological samples from all categories of bone tumors and integrated multiple neural network architectures for machine learning models. We then evaluated the accuracy of these models in diagnosing primary bone tumors. MethodsIn this paper, the machine learning model based on the deep convolutional neural network (DC-NN) method was combined with imaging omics analysis to analyze and discuss its clinical value in diagnosing primary bone tumors. In addition, this paper proposed a screening method for differentially expressed genes. Based on the paired T-test method, the process first estimated the tumor purity in the experimental data of each sample case, then assessed the actual gene expression value of the experimental data of each sample case, and finally calculated the optimized paired T-test statistics, and screened differentially expressed genes according to the threshold value. ResultsThe selected model demonstrated excellent diagnostic accuracy in distinguishing between normal and tumor images, with overall accuracy of (99.8 ± 0.4) % for five rounds of testing using the DCNN model and positive and negative predictive values of (100.0 ± 0.0) % and (99.6 ± 0.8) %, respectively. The mean area under each dataset’s curve (AUC) was (0.998 ± 0.004). Further, ten rounds of testing using the DCNN model showed an overall accuracy of (71.2 ± 1.6) % and a substantial positive predictive value of (91.9 ± 8.5) % in distinguishing benign from malignant bone tumors, with an average AUC of (0.62 ± 0.06) across datasets. ConclusionThe deep learning model accurately classifies bone tumor histopathology based on the degree of infiltration, achieving diagnostic performance comparable to that of senior pathologists. These findings affirm the feasibility and effectiveness of histopathological diagnosis in bone tumors, providing a theoretical foundation for the application and advancement of machine learning-assisted histopathological diagnosis in this field.

Rapid and noncontact identification of soybean flour in edible insect using NIR spectral imager: A case study in Protaetia brevitarsis seulensis powder

Edible insects are notably considered novel foods with high amounts of protein, making them valuable. There are still no reported cases of edible insect adulteration, but there is a potential issue as valuable products, particularly during supply chains. This work demonstrated the feasibility of near-infrared hyperspectral imaging (NIR-HSI), ranging from 1000 nm to 2100 nm, for rapid and nondestructive identification of soybean flour in Protaetia brevitarsis seulensis (PBS) powder. Three different approaches to soybean flour detection were realized by using an extended principal component analysis (PCA), data-driven-soft independent modelling of class analogy (DD-SIMCA), and regression algorithms, namely partial least squares regression (PLSR) and one-dimensional convolutional neural networks (1D-CNN). Our study demonstrated that extended PCA for soybean flour pixel identification showed a poor linear correlation (R2 = 0.835) and the error (RMSE = 12.39%) between the identified soybean flour pixel and its actual concentrations. By employing DD-SIMCA, 100% accuracy was achieved, allowing the superior performance of one-class classification method. In conjunction with regression methods, 1D-CNN with the Savitzky-Golay first derivative (SG1) spectra generated the optimum prediction accuracy, indicated by an R2P of 0.99, an RMSEP of 1.15%, and an RPD of 12.92. Furthermore, a chemical image derived from the 1D-CNN showed a clear visualization of adulterated PBS. Finally, NIR-HSI optimized with a 1D-CNN model could be a promising technique for the identification of soybean flour in PBS powder in a nondestructive manner.

Online and Offline Fault Detection and Diagnostics in a Nuclear Power Plant Condenser

Nuclear power plants (NPPs) face significant financial pressures due to operational and maintenance costs. This research investigates Fault Detection and Diagnostics (FDD) techniques to optimize maintenance schedules and reduce expenses. The NPP condenser plays a critical role in converting turbine exhaust steam back into water for reuse. Condenser tube fouling, a prevalent fault mode, impedes heat transfer efficiency and can lead to decreased plant efficiency and safety risks. This study proposes an FDD framework that leverages raw signal analyses from temperature and pressure monitoring to detect and diagnose condenser tube fouling in both online and offline settings. The online approach facilitates close-to-real-time predictions, enabling proactive maintenance strategies. Additionally, the framework explores incorporating a condenser’s maintenance history for enhanced diagnostics. We employ a dataset obtained from a simulated nuclear power plant condenser using the Asherah Nuclear Power Plant Simulator (ANS). ANS replicates the operational dynamics of a pressurized water reactor (PWR) type NPP. The proposed methodology utilizes an encoder-decoder (E-D) structured 1DCNN model to analyze the raw signals. The research demonstrates consistent and accurate fault detection and diagnostics for condenser tube fouling in both online and offline scenarios. A high potential for generalization to unseen conditions was observed. However, online detection using small data windows necessitates caution due to potential false alarms around the transition points. Our findings pave the way for further exploration of robust diagnostics by accommodating a wider spectrum of fouling rates within degradation classes using ANS. This combined online and offline FDD approach offers a promising solution for promoting operational safety, efficiency, and cost-effectiveness in NPP condensers.

Deep learning-assisted local resonance strategy for accurate internal damage imaging in composites

In this paper, we propose a deep neural network-assisted strategy to accurately and efficiently identify local defect resonance (LDR) modes and accurately image the internal damage in composites. A two-dimensional convolutional neural network (2D-CNN) model was constructed to identify LDR modes. The frequency-domain contour maps were used as input data, given that the LDR phenomenon exhibits discernible physical attributes in the frequency domain that are conducive to deep neural network assimilation. The obtained results demonstrate effective training outcomes and transferability, even with a limited number of samples. The LDR modes are efficiently extracted by the developed 2D-CNN model and used to obtain the accurate imaging of internal damages in composites.

Leveraging 3D Convolutional Neural Networks for Accurate Recognition and Localization of Ankle Fractures.

Ankle fractures are common injuries with substantial implications for patient mobility and quality of life. Traditional imaging methods, while standard, have limitations in detecting subtle fractures and distinguishing them from complex bone structures. The advent of 3D Convolutional Neural Networks (3D-CNNs) offers a promising avenue for enhancing the accuracy and reliability of ankle fracture diagnoses. In this study, we acquired 1453high-resolution CT scans and processed them through three distinct 3D-CNN models: 3D-Mobilenet, 3D-Resnet101, and 3D-EfficientNetB7. Our approach involvedmeticulous preprocessing of images, including normalization and resampling, followed by a systematic comparative evaluation of the modelsbased on accuracy, Area Under the Curve (AUC), and recall metrics. Additionally, the integration of Gradient-weighted Class Activation Mapping (Grad-CAM) provided visual interpretability of the models' predictive focus points. The 3D-EfficientNetB7 model outperformed the other models, achieving an accuracy of 0.91 and an AUC of 0.94 after 20 training epochs. It demonstrated particularly effective in the accurate detection and localization of subtle and complex fractures. Grad-CAM visualizations confirmed the model's focus on clinically relevant areas, aligning with expert assessments and enhancing trust in automated diagnostics. Spatial localization techniques were pivotal in improving interpretability, offering clear visual guidance for pinpointing fracture sites. Our findings highlight the effectiveness of the 3D-EfficientNetB7 model in diagnosing ankle fractures, supported by robust performance metrics and enhanced visualization tools.

A 3D CNN Model Framework for Early Identification and Classification of Alzheimer’s in Brain MRI Images

A 3D CNN Model Framework for Early Identification and Classification of Alzheimer’s in Brain MRI Images

MosquitoSong+: A noise-robust deep learning model for mosquito classification from wingbeat sounds.

In order to assess risk of mosquito-vector borne disease and to effectively target and monitor vector control efforts, accurate information about mosquito vector population densities is needed. The traditional and still most common approach to this involves the use of traps along with manual counting and classification of mosquito species, but the costly and labor-intensive nature of this approach limits its widespread use. Numerous previous studies have sought to address this problem by developing machine learning models to automatically identify species and sex of mosquitoes based on their wingbeat sounds. Yet little work has addressed the issue of robust classification in the presence of environmental background noise, which is essential to making the approach practical. In this paper, we propose a new deep learning model, MosquitoSong+, to identify the species and sex of mosquitoes from raw wingbeat sounds so that it is robust to the environmental noise and the relative volume of the mosquito's flight tone. The proposed model extends the existing 1D-CNN model by adjusting its architecture and introducing two data augmentation techniques during model training: noise augmentation and wingbeat volume variation. Experiments show that the new model has very good generalizability, with species classification accuracy above 80% on several wingbeat datasets with various background noise. It also has an accuracy of 93.3% for species and sex classification on wingbeat sounds overlaid with various background noises. These results suggest that the proposed approach may be a practical means to develop classification models that can perform well in the field.

How does lead selection impact diagnosis through AI? Implications for continuous monitoring systems using convolutional neural networks

Abstract Background and Objectives Recently, there has been a surge in non-invasive diagnosing methods leveraging the dipolar 12-lead ECG information using Convolutional Neural Networks (CNNs). However, the traditional ECG contains 12 projections of the cardiac dipole in a three-dimensional space can result in redundancy. Furthermore, in continuous monitoring systems such as Holter and wearables, not all 12 leads are consistently available. This study aims to quantify redundancy and determine the best lead configuration for improved classification results using a CNN. It evaluates the level of redundancy in ECG leads and its impact on CNN classification. Methods The degree of redundancy was quantified by means of lead entropies (H) as depicted in Figure 1.A. The standard 12-lead ECG was chosen as the baseline input. From it, various subsets of leads were selected (6, 3, 1 leads) with the objective of attaining maximal orthogonality between leads while minimizing redundancy (Figure 1.B). A 2D-CNN model was trained for multiclass classification: Hypertension (HYP), ST-T interval Changes (STTC), Conduction Disturbances (CD), Myocardial Infarction (MI) and Normal (NORM). The designed inputs were employed to evaluate how redundancy impacts the performance of the CNN (see Figure 1.C). Results Reducing redundancy in input leads led to notable improvements in model prediction for all pathological conditions in terms of Area Under the Curve (AUC). Reducing to 6 and 8 leads were the most suitable input for the CNN (Table 1) increasing by 1.18% in HYP, 0.23% in STTC, 0.52% in CD, and 0.33% in MI. However, a significant reduction in redundancy may result in the loss of essential clinical information for classification, as observed in scenarios involving only 3 or 1 leads. Conclusions When diagnosing pathologies with systems featuring fewer leads, its combinations should aim for maximum orthogonality to reduce redundancy while preserving unique information. By strategically selecting a set of leads with reduced input redundancy, the effectiveness of the standard 12-lead ECG can be improved, thereby enhancing the utility of such continuous monitoring systems for automated diagnosis and also decreasing training and prediction times.

HepScope: CNN-based single-cell discrimination of malignant hepatocytes

Hepatocellular carcinoma (HCC) presents a major health issue worldwide. This study introduces the HepScope gene set, developed through a hybrid approach that utilizes single-cell RNA sequencing (scRNA-seq), spatial transcriptomics (stRNA-seq), bulk RNA-seq, and proteomics data, along with the adaptation of Convolutional Neural Network (CNN) techniques. Comprising 113 genes significantly upregulated in malignant hepatocytes, HepScope gene set includes 77 genes unique to this set compared to other HCC-related gene sets. Unlike existing solutions, HepScope gene set demonstrated superior discriminatory power in distinguishing malignant from non-malignant hepatocytes, as validated by Seurat's module score, outperforming five other gene sets across multiple datasets. A 1D-CNN model, specifically adapted for the HepScope gene set, achieved superior accuracy (0.71), AUROC (0.82), and F1 score (0.85) compared to models trained on other gene sets, underscoring its enhanced predictive precision and robustness. Rigorous cross-validation and benchmarking against alternative models confirmed HepScope’s consistent performance. Additionally, we evaluated the prognostic capability of the HepScope gene set across two independent cohorts, demonstrating that the HepScope-based risk score is a strong independent predictor of overall and disease-free survival, making it a valuable tool for patient prognosis. Collectively, our findings position HepScope as a promising gene set for both diagnostic and prognostic applications in HCC, highlighting its potential in precision medicine for HCC and offering insights into personalized therapy development.

Towards the Development of the Clinical Decision Support System for the Identification of Respiration Diseases via Lung Sound Classification Using 1D-CNN.

Respiratory disorders are commonly regarded as complex disorders to diagnose due to their multi-factorial nature, encompassing the interplay between hereditary variables, comorbidities, environmental exposures, and therapies, among other contributing factors. This study presents a Clinical Decision Support System (CDSS) for the early detection of respiratory disorders using a one-dimensional convolutional neural network (1D-CNN) model. The ICBHI 2017 Breathing Sound Database, which contains samples of different breathing sounds, was used in this research. During pre-processing, audio clips were resampled to a uniform rate, and breathing cycles were segmented into individual instances of the lung sound. A One-Dimensional Convolutional Neural Network (1D-CNN) consisting of convolutional layers, max pooling layers, dropout layers, and fully connected layers, was designed to classify the processed clips into four categories: normal, crackles, wheezes, and combined crackles and wheezes. To address class imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) was applied to the training data. Hyperparameters were optimized using grid search with k-fold cross-validation. The model achieved an overall accuracy of 0.95, outperforming state-of-the-art methods. Particularly, the normal and crackles categories attained the highest F1-scores of 0.97 and 0.95, respectively. The model's robustness was further validated through 5-fold and 10-fold cross-validation experiments. This research highlighted an essential aspect of diagnosing lung sounds through artificial intelligence and utilized the 1D-CNN to classify lung sounds accurately. The proposed advancement of technology shall enable medical care practitioners to diagnose lung disorders in an improved manner, leading to better patient care.

Enhancing Multimodal Emotional Information Extraction in Film and Television through Adaptive Feature Fusion with DenseNe, Transformer, and 3D CNN Models

ABSTRACT The extraction of multimodal emotional information enables a more nuanced representation of the emotional subtleties embedded in film and television works. However, conventional approaches that independently extract features from images and text fail to capture the rich semantic interplay between these modalities, impeding the feature learning process within each modality. To address this, this paper presents an innovative model for extracting emotional information from multimodal film and television content. The model utilizes DenseNe for image feature extraction, enhancing network depth via the MSC module. Text feature extraction is achieved through a Transformer encoder, while video feature extraction employs a 3D CNN model. Refinements are made to the number and placement of convolutional layers, planar convolution size, and 3D convolution depth. Moreover, a multi-head scaled dot-product attention mechanism is incorporated into the interaction module to compute the similarity between each image block within the sequence and every word in the text sequence. Experimental evaluations on the CMU-MOSEI and CMU-MOSI datasets showcase superior performance compared to the baseline model, achieving accuracy and F1 scores of 0.708 and 0.698, respectively. Noteworthy is the proposed adaptive feature fusion module, which enriches the expressiveness of pivotal emotional features while eliminating redundant data.

Enhancing transparency and fairness in automated credit decisions: an explainable novel hybrid machine learning approach

This paper uses a generalised stacking method to introduce a novel hybrid model that combines a one-dimensional convolutional neural network 1DCNN with extreme gradient boosting XGBoost. We compared the predictive accuracies of the proposed hybrid architecture with three conventional algorithms-1DCNN, XGBoost and logistic regression (LR) using a dataset of over twenty thousand peer-to-peer (P2P) consumer credit observations. By leveraging the SHAP algorithm, the research provides a detailed analysis of feature importance, contributing to the model’s predictions and offering insights into the overall and individual significance of different features. The findings demonstrate that the hybrid model outperforms the LR, XGBoost and 1DCNN models in terms of classification accuracy. Furthermore, the research addresses concern regarding fairness and bias by showing that removing potentially discriminatory features, such as age and gender, does not significantly impact the hybrid model’s classification capabilities. This suggests that fair and unbiased credit scoring models can achieve high effectiveness levels without compromising accuracy. This paper makes significant contributions to academic research and practical applications in credit risk management by presenting a hybrid model that offers superior classification accuracy and promotes interpretability using the model agnostic SHAP framework.

Analyzing TVB-N in snakehead by Bayesian-optimized 1D-CNN using molecular vibrational spectroscopic techniques: Near-infrared and Raman spectroscopy

Total volatile basic nitrogen (TVB-N) is one of the key indicators for assessing fish freshness. This research employed near-infrared (NIR) and Raman spectroscopy methods to detect the TVB-N content in snakehead fillets. We extracted feature variables associated with TVB-N from NIR and Raman spectroscopy using Variable Crossover Point Arithmetic - Improved Reduced-Input Vector (VCPA-IRIV). Using these features, we established partial least squares (PLS) and One-dimensional Convolutional Neural Network (1D-CNN) models. Subsequently, data fusion strategies were employed to predict the TVB-N content. Notably, feature-level fusion in conjunction with Bayesian-optimized 1D-CNN, reached the best results, as evidenced by calibration and predictive correlation coefficients of 0.9677 and 0.9676 for TVB-N. These findings underscore the effectiveness of both NIR and Raman spectroscopy in evaluating fish freshness. The fusion of these two vibrational spectroscopy techniques enables a more rapid, efficient and comprehensive quantification of fish freshness.

Bayesian optimization of one-dimensional convolutional neural networks (1D CNN) for early diagnosis of Autistic Spectrum Disorder

Autistic Spectrum Disorder (ASD) is a challenging neurological development disorder, which involves poor social interaction, communication, and repetitive behaviours. If autism is identified early enough it can be treated with better outcomes but present diagnostic tests are dependent on subjective opinion, consume a lot of time, and are vague. This study is aimed at optimizing one-dimensional convolutional neural networks (1D CNN) to improve the precision and speed of early ASD diagnosis. Four ASD datasets representing different age groups — toddlers, children, adolescents, and adults were modelled using one-dimensional convolutional neural networks (1D CNN). These datasets are accessible to the public on the UCI Machine Learning Repository and Kaggle, they consist of behavioural features relevant to ASD diagnosis. Each dataset underwent feature selection, categorical encoding, and missing value handling. Then, baseline 1D CNN with predefined hyperparameters was modelled on each of the datasets. Subsequently, the baseline models were optimized using the Tree-structured Parzen Estimator (TPE). An interactive web-based ASD diagnostic tool was developed, where user inputs are processed through age-specific pre-trained optimized models to determine ASD probability. The optimized 1D CNN models significantly outperformed the baseline models across all age groups and achieved scores of 100% in accuracy, precision, recall, F1-score, MCC, and AUC ROC. This implies that the optimized models can reliably identify people in various age groups who have and do not have ASD. The development of an interactive web-based diagnostic tool extends the practical utility of the models, making them accessible for clinical and at-home use.

Deep learning-based anomaly detection using one-dimensional convolutional neural networks (1D CNN) in machine centers (MCT) and computer numerical control (CNC) machines.

Computer numerical control (CNC) and machine center (MCT) machines are mechanical devices that manipulate different tools using computer programming as inputs. Predicting failures in CNC and MCT machines before their actual failure time is crucial to reduce maintenance costs and increase productivity. This study is centered around a novel deep learning-based model using a 1D convolutional neural network (CNN) for early fault detection in MCT machines. We collected sensor-based data from CNC/MCT machines and applied various preprocessing techniques to prepare the dataset. Our experimental results demonstrate that the 1D-CNN model achieves a higher accuracy of 91.57% compared to traditional machine learning classifiers and other deep learning models, including Random Forest (RF) at 89.71%, multi-layer perceptron (MLP) at 87.45%, XGBoost at 89.67%, logistic regression (LR) at 75.93%, support vector machine (SVM) at 75.96%, K-nearest neighbors (KNN) at 82.93%, decision tree at 88.36%, naïve Bayes at 68.31%, long short-term memory (LSTM) at 90.80%, and a hybrid 1D CNN + LSTM model at 88.51%. Moreover, our proposed 1D CNN model outperformed all other mentioned models in precision, recall, and F-1 scores, with 91.87%, 91.57%, and 91.63%, respectively. These findings highlight the efficacy of the 1D CNN model in providing optimal performance with an MCT machine's dataset, making it particularly suitable for small manufacturing companies seeking to automate early fault detection and classification in CNC and MCT machines. This approach enhances productivity and aids in proactive maintenance and safety measures, demonstrating its potential to revolutionize the manufacturing industry.

A fast hybrid approach for continuous land cover change monitoring and semantic segmentation using satellite time series

Land cover change detection and classification, including both inter-class changes (land cover conversion, LCC) and intra-class changes (land cover modification, LCM), is critical for understanding the Earth’s dynamic processes and promoting sustainability. However, previous studies have predominantly focused on LCC, with less emphasis on LCM. Land cover classification remains challenging, and its mapping results are often affected by salt and pepper noise. Here, we propose a hybrid approach for continuous change detection and classification of LCC and LCM using Jinchang City, China, as a case study. Firstly, we combined the Continuous Change Detection and Classification (CCDC) and the Bayesian Estimator of Abrupt change, Seasonal change, and Trend (BEAST) algorithms to identify LCC and LCM using all available Landsat time series (TS) data from 2000 to 2020. Then, the harmonic regression coefficients and RMSE values derived from CCDC (hereafter called CCDC features) were fed into the DCNN model for LCC classification. Our findings indicate: (1) For LCC and LCM accuracy assessment, the CCDC and BEAST ensemble achieved a spatial F1 score of 82.7% and an average temporal F1 score of 79.7%. (2) In LCC classification, the DCNN model with CCDC features, particularly DeepLabV3+, outperformed the pixel-based XGBoost and other multi-year land cover products, with frequency-weighted intersection over union (FWIoU), overall accuracy, and Kappa scores of 88.7%, 94%, and 0.87, respectively. (3) Seasonal LCM showed a more concentrated distribution than trend LCM. (4) In Jinchang City, LCM larger than LCC in area, and grassland and cultivated land are the most distributed. Our approach can be contributed to wall-to-wall land surface monitoring and enhance land management capabilities.