AE Model Research Articles

Research Progress on Gene Regulation of Plant Floral Organogenesis.

Flowers, serving as the reproductive structures of angiosperms, perform an integral role in plant biology and are fundamental to understanding plant evolution and taxonomy. The growth and organogenesis of flowers are driven by numerous factors, such as external environmental conditions and internal physiological processes, resulting in diverse traits across species or even within the same species. Among these factors, genes play a central role, governing the entire developmental process. The regulation of floral genesis by these genes has become a significant focus of research. In the AE model of floral development, the five structural whorls (calyx, corolla, stamens, pistils, and ovules) are controlled by five groups of genes: A, B, C, D, and E. These genes interact to give rise to a complex control system that governs the floral organsgenesis. The activation or suppression of specific gene categories results in structural modifications to floral organs, with variations observed across different species. The present article examines the regulatory roles of key genes, including genes within the MADS-box and AP2/ERF gene clusters, such as AP1, AP2, AP3, AG, STK, SHP, SEP, PI, and AGL6, as well as other genes, like NAP, SPL, TGA, PAN, and WOX, in shaping floral organ genesis. In addition, it analyzes the molecular-level effects of these genes on floral organ formation. The findings offer a deeper understanding of the genetic governance of floral organ genesis across plant species.

Coal–Rock Catastrophic Collapse: Precursors Based on AE and Fiber Bundle Models

Coal–Rock Catastrophic Collapse: Precursors Based on AE and Fiber Bundle Models

CL‐20/4,5‐MDNI crystal growth and morphology under different conditions: A molecular dynamics simulation

AbstractCL‐20/4,5‐MDNI exists in two cocrystal forms with different stoichiometric ratios of 1:1 and 1:3. The properties (such as blast performance, thermal performance, sensitivity, and density) of the two cocrystals are significantly different, but both cocrystals are expected to become new high‐energy insensitive explosives and have good applications. The AE model and molecular dynamics were used to simulate the crystal morphology and important crystal planes of 1:1 and 1:3 CL‐20/4,5‐MDNI under vacuum conditions, as well as in five solvents and different temperatures. 1:1 CL‐20/4,5‐MDNI has five main growth crystal planes of (002), (102), (111), (020) and (021), of which (002) has the largest proportion. 1:3 CL‐20/4,5‐MDNI has six main growth crystal planes of (001), (010), (011), (100), (101) and (111), of which (001) accounts for the largest proportion. The crystal morphology aspect ratio of 1:3 CL‐20/4,5‐MDNI in acetonitrile and 1:1 CL‐20/4,5‐MDNI in methanol are 1.936 and 1.680, respectively. The results were consistent with the experiments, which has certain guiding significance for CL‐20/4,5‐MDNI to obtain a crystal morphology closer to spherical shape.

Anomaly Detection Method for Harmonic Reducers with Only Healthy Data.

A harmonic reducer is an important component of industrial robots. In practical applications, it is difficult to obtain enough anomaly data from error cases for the supervised training of models. Whether the information contained in regular features is sensitive to anomaly detection is unknown. In this paper, we propose an anomaly detection frame for a harmonic reducer with only healthy data. We considered an auto-encoder trained using only healthy features, such as feature mapping, in which the difference between the output and the input constitutes a new high-dimensional feature space that retained information relevant only to anomalies. Compared to the original feature space, this space was more sensitive to abnormal data. The mapped features were then fed into the OCSVM to preserve the feature details of the abnormal information. The effectiveness of this method was validated by multiple sets of data collecting from harmonic reducers. Three different residual calculations and four different AE models were used, showing that the method outperforms an AE or an OCSVM alone. It is also verified that the method outperforms other typical anomaly detection methods.

An efficient deep learning prognostic model for remaining useful life estimation of high speed CNC milling machine cutters

CNC machines are engaged in numerous industries, including critical ones like the aerospace, automotive, and military sectors, among others. Sensor data are time-series that may suffer from complex interconnections between variables and dynamic features. Long Short Term Memory LSTM excels in dynamic feature extraction, and Autoencoder AE has great capabilities in nonlinear deep knowledge of time-series data variables. In this work, we propose a model for tool wear prediction of CNC milling machine cutters as a type of time-series data taking advantage of the LSTM and AE capabilities. The framework consists of many steps, including extracting multi-domain features and a correlation analysis to select the most correlated features to the tool wear. New features are added, such as entropy and interquartile range IQR, which proved to be highly correlated to the cutter tool wear. An LSTM`-AE model is then trained, validated, and tested on this feature map to predict the target tool wear value. The model is provided with degradation or Run-To-Failure data for CNC machine cutters, the PHM10 dataset, to predict the tool wear values. The predicted tool wear value is compared against the wear curve to estimate RUL values. The predicted RUL values mostly underestimate the real values, which helps schedule for maintenance or equipment replacement before failure. The experimental results show that the proposed framework outperforms state-of-the-art DL methods in tool wear prediction accuracy approaching %98, as well as an enhancement of MAE and RMSE in the test set by reaching 2.6 ± 0.3222E-3 and 3.1 ± 0.6146 E-3, respectively.

Specific and Polyfunctional T Cell Response Against N-Methyl-d-aspartate Receptor in an Autoantibody-Mediated Encephalitis Model.

Anti-N-Methyl-d-aspartate receptor (NMDAR) autoimmune encephalitis (NMDAR AE) is an autoimmune disease characterized by severe psychiatric and neurological symptoms. While the pathogenic role of antibodies (Abs) directed against the GluN1 subunit of NMDAR is well described in this disease, the immune mechanisms involved in the generation of the autoimmune B cell response, especially the role of T helper cells, are poorly understood. Previously, we developed a B-cell-mediated mouse model of NMDAR AE by immunization with a GluN1359-378 peptide that drives a series of symptoms that recapitulate AE such as anxiety behaviour and spatial memory impairment. In this mouse model, we identified anti-GluN1-specific CD4+ but also CD8+ T cells in both spleen and meninges. T helper cells have a polyfunctional profile, arguing for a T and B cell crosstalk to generate anti-GluN1 pathogenic Abs. Interestingly, proteomic analysis of AE meninges showed enrichment of differentially expressed proteins in biological processes associated with B cell activation and cytokine signalling pathways. This study identified, for the first time, a potential contribution of T helper cells in the pathology of NMDAR AE and paved the way for the development of future tolerogenic approaches to treat relapses.

Long Short-Term Memory Autoencoder for Anomaly Detection in Rails Using Laser Doppler Vibrometer Measurements

Abstract This study presents an application of Long Short-Term Memory Autoencoder (LSTM AE) for the detection of broken rails based on laser Doppler Vibrometer (LDV) measurements. This work is part of an ongoing project aimed at developing a non-contact damage detection system using LDV measurements. The damage detection system consists of two laser Doppler vibrometers (LDV) mounted on a moving rail car to measure vibrations induced on the rail head. Field tests were carried out at the Transportation Technology Center (TTC) in Pueblo, CO, to collect the vibrational data. This study focused on the detection of broken rails. To simulate the reflected and transmitted waves induced by the broken rail, a welded joint was used. The data was collected from moving LDV measurements, in which the train was operating at three different speeds: 16km/h (10mph), 32km/h (20mph), and 48km/h (30mph). After obtaining the data, filtering and signal processing were applied to obtain the signal features in time and frequency domains. Next, correlation analysis and principal component analysis were carried out for feature selection and dimension reduction to determine the input used to train and test the LSTM AE model. In this study, the LSTM AE models were trained based on different data sets for anomaly detection. Consequently, an automatic anomaly detection approach for anomaly detection based on the LSTM AE model was evaluated. The results show that the LSTM AE model can efficiently detect the anomaly based on the selected features at three different speeds.

Genetic and environmental contributions to the development of soft tissue facial profile: a twin study.

This study aimed to determine the relative contribution of genetic and environmental factors in the phenotypic variation of the soft tissue facial profile during the mixed dentition and the permanent dentition stages. In this retrospective cohort study, standardized facial profile photographs of 139 twin pairs (55 monozygotic and 84 dizygotic) were obtained from archival records at the Adelaide Dental School. Photographic analysis used 12 angular and 14 linear facial profile measurements from the mixed dentition (7-11 years) to the permanent dentition (12-17 years) stages. A genetic analysis was performed using a univariate structural equation model adhering to the normal assumptions of a twin model. In the mixed dentition stage, the additive genetic (A) and unique environment (E) model, AE model, was the most parsimonious in explaining the observed phenotypic variance for all 26 facial traits with the narrow-sense heritability estimates ranging between 0.38 and 0.79. In the permanent dentition, the AE model was the most parsimonious for 20 out of 26 traits, however, the variance of six traits, particularly those in the lower third of the face, was best explained by the shared environmental and unique environmental factors. This study exclusively included twins of European ancestry. The soft tissue facial profile demonstrated dynamic genetic and environmental influences with a greater additive genetic influence during the mixed dentition and the early stages of the permanent dentition. However, there was evidence of increasing environmental influence in the lower third of the face during the early stages of the permanent dentition.

Tilts, developmental modules, and cognitive differentiation-integration effort: A multi-study response to

Sorjonen et al. (2024) critique a recently published finding that cognitive tilts are heritable, which was advanced as a line of evidence supporting their substantive (rather than artefactual) nature. These researchers claim: i) that the heritability of tilts is simply a function of the heritabilities of the specific cognitive dimensions used in their estimation, and ii) that spuriously heritable tilts can be recovered using difference scores between psychometric, anthropometric, and even random number variables. Here, multiple studies employing three behavior genetic datasets are used to test these claims. Even when cognitive tilts are residualized for their association with their constituent abilities, they still exhibit small, but non-zero heritabilities. Shared environmentality (C) accounts for the largest proportion of variance among these residuals. Tilts generated using random numbers are, by contrast, in all cases associated with AE models, exhibiting near 100 % E variance, corresponding to error. In the Swedish Twin Registry, the tilt residual is positively correlated with a measure of life history speed (Mini-K score), suggesting that tilts capture cognitive differentiation-integration effort conditioned developmentally by C variance. Distinct latent factors among psychometric and anthropometric variables in the Georgia Twin Study are also found. These indicate the presence of distinct developmental modules, meaning that tilts estimated using manifest variables associated with different modules lack theoretical credibility, as also evidenced by weak cross loadings.

Structural Damage Identification Using Autoencoders: A Comparative Study

Structural health monitoring (SHM) ensures the safety and reliability of civil infrastructure. Autoencoders, as unsupervised learning models, offer promise for SHM by learning data features and reducing dimensionality. However, comprehensive studies comparing autoencoder models in SHM are scarce. This study investigates the effectiveness of four autoencoder-based methodologies, combined with Hotelling’s T2 statistical tool, to detect and quantify structural changes in three civil engineering structures. The methodologies are evaluated based on computational costs and their abilities to identify structural anomalies accurately. Signals from the structures, collected by accelerometers, feed the autoencoders for unsupervised classification. The latent layer values of the autoencoders are used as parameters in Hotelling’s T2, and results are compared between classes to assess structural changes. Average execution times of each model were calculated for computational efficiency. Despite variations, computational cost did not hinder any methodology. The study demonstrates that the best fitting model, VAE-T2, outperforms its counterparts in identifying and quantifying structural changes. While the AE, SAE, and CAE models showed limitations in quantifying changes, they remain relevant for detecting anomalies. Continuous application and development of these techniques contribute to SHM advancements, enabling the increased safety, cost-effectiveness, and long-term durability of civil engineering structures.

Genetic and environmental determinants of bone quality: a cross-sectional analysis of the Hungarian Twin Registry.

There is abundant evidence that bone mineral content is highly heritable, while the heritability of bone quality (i.e. trabecular bone score [TBS] and quantitative ultrasound index [QUI]) is rarely investigated. We aimed to disentangle the role of genetic, shared and unique environmental factors on TBS and QUI among Hungarian twins. Our study includes 82 twin (48 monozygotic, 33 same-sex dizygotic) pairs from the Hungarian Twin Registry. TBS was determined by DXA, QUI by calcaneal bone ultrasound. To estimate the genetic and environmental effects, we utilized ACE-variance decomposition. For the unadjusted model of TBS, an AE model provided the best fit with > 80% additive genetic heritability. Adjustment for age, sex, BMI and smoking status improved model fit with 48.0% of total variance explained by independent variables. Furthermore, there was a strong dominant genetic effect (73.7%). In contrast, unadjusted and adjusted models for QUI showed an AE structure. Adjustments improved model fit and 25.7% of the total variance was explained by independent variables. Altogether 70-90% of the variance in QUI was related to additive genetic influences. We found a strong genetic heritability of bone quality in unadjusted models. Half of the variance of TBS was explained by age, sex and BMI. Furthermore, the adjusted model suggested that the genetic component of TBS could be dominant or an epistasis could be present. In contrast, independent variables explained only a quarter of the variance of QUI and the additive heritability explained more than half of all the variance.

Anomaly detection of four-engine aircraft based on Parameter differences

Abstract In this paper, a real-time anomaly detection method combining GDN and Gaussian model is proposed for the real-time engine condition monitoring of four-engine aircraft. Firstly, the gas path parameters of the engine are selected and the difference of the gas path parameters is calculated to offset the influence of the external environment on the engine. Thirdly, the sliding time window method is used for real-time prediction, and the model prediction residuals are used for unsupervised anomaly detection. Finally, the Gaussian model is used to prune the anomalies detected by the GDN model to reduce the misjudgment of anomalies and improve the accuracy of anomaly detection. The experimental results show that the proposed method can detect 97.96% of the outliers in the test set and the accuracy reaches 89.47%, which is better than LSTM, AE, GDN, and Gaussian models alone.

A Prediction of Heart Disease using IoT based ThingSpeak Basis and Deep Learning Method

Developing an IoT-based system with enhanced deep learning to predict heart disease aims to revolutionize healthcare. The system, by continuously monitoring vital signs, utilizes advanced machine learning for early detection of potential heart issues, emphasizing innovation for timely interventions and potentially life-saving outcomes. Select appropriate sensors for monitoring vital signs such as heart rate, blood pressure, ECG (Electrocardiogram), oxygen saturation, and temperature. Connect these sensors to an IoT platform using wireless communication protocols like Wi-Fi. Predicting heart disease using an IoT-based ThingSpeak framework is a practical and efficient way to monitor and predict heart-related issues in real-time. ThingSpeak is an IoT platform that allows you to collect, store, analyse, and visualize data from IoT sensors. Here's a step-by-step guide on how to develop a heart disease prediction system using ThingSpeak: Deploy the trained deep learning model to make real-time predictions based on the IoT sensor data. Implement strong security regulations. The input dataset is pre-processed approach for resolving error data and missing values. On ThingSpeak, we can apply preprocessing functions to your data. This might include data smoothing, outlier removal, or missing data handling. Pre-processed data values features are then classified into normal and abnormal by Convolution Neural Network-CNN. A 10-fold cross-validation tactic was employed during the model development procedure. In the experimental result comparisons of MCC calculation, the MD2N model attained the 0.86201 and DCNN perfect reached as 0.84111 and then DBN reached as 0.91157 and then AE model rate as 0.88662 and then proposed model reaches as 0.93291 correspondingly.

Prediction of Adolescents' Fluid Intelligence Scores based on Deep Learning with Reconstruction Regularization.

The aim of this study was to develop a predictive model for uncorrected/actual fluid intelligence scores in 9-10 year old children using magnetic resonance T1-weighted imaging. Explore the predictive performance of an autoencoder model based on reconstruction regularization for fluid intelligence in adolescents. We collected actual fluid intelligence scores and T1-weighted MRIs of 11,534 adolescents who completed baseline tasks from ABCD Data Release 3.0. A total of 148 ROIs were selected and 604 features were proposed by FreeSurfer segmentation. The training and testing sets were divided in a ratio of 7:3. To predict fluid intelligence scores, we used AE, MLP and classic machine learning models, and compared their performance on the test set. In addition, we explored their performance across gender subpopulations. Moreover, we evaluated the importance of features using the SHapley Additive Explain method. Results: The proposed model achieves optimal performance on the test set for predicting actual fluid intelligence scores (PCC = 0.209 ± 0.02, MSE = 105.212 ± 2.53). Results show that autoencoders with refactoring regularization are significantly more effective than MLPs and classical machine learning models. In addition, all models performed better on female adolescents than on male adolescents. Further analysis of relevant characteristics in different populations revealed that this may be related to gender differences in underlying fluid intelligence mechanisms. We construct a weak but stable correlation between brain structural features and raw fluid intelligence using autoencoders. Future research may need to explore ensemble regression strategies utilizing multiple machine learning algorithms on multimodal data in order to improve the predictive performance of fluid intelligence based on neuroimaging features.

Knowledge-based quality assurance of a comprehensive set of organ at risk contours for head and neck radiotherapy.

Manual review of organ at risk (OAR) contours is crucial for creating safe radiotherapy plans but can be time-consuming and error prone. Statistical and deep learning models show the potential to automatically detect improper contours by identifying outliers using large sets of acceptable data (knowledge-based outlier detection) and may be able to assist human reviewers during review of OAR contours. This study developed an automated knowledge-based outlier detection method and assessed its ability to detect erroneous contours for all common head and neck (HN) OAR types used clinically at our institution. We utilized 490 accurate CT-based HN structure sets from unique patients, each with forty-two HN OAR contours when anatomically present. The structure sets were distributed as 80% for training, 10% for validation, and 10% for testing. In addition, 190 and 37 simulated contours containing errors were added to the validation and test sets, respectively. Single-contour features, including location, shape, orientation, volume, and CT number, were used to train three single-contour feature models (z-score, Mahalanobis distance [MD], and autoencoder [AE]). Additionally, a novel contour-to-contour relationship (CCR) model was trained using the minimum distance and volumetric overlap between pairs of OAR contours to quantify overlap and separation. Inferences from single-contour feature models were combined with the CCR model inferences and inferences evaluating the number of disconnected parts in a single contour and then compared. In the test dataset, before combination with the CCR model, the area under the curve values were 0.922/0.939/0.939 for the z-score, MD, and AE models respectively for all contours. After combination with CCR model inferences, the z-score, MD, and AE had sensitivities of 0.838/0.892/0.865, specificities of 0.922/0.907/0.887, and balanced accuracies (BA) of 0.880/0.900/0.876 respectively. In the validation dataset, with similar overall performance and no signs of overfitting, model performance for individual OAR types was assessed. The combined AE model demonstrated minimum, median, and maximum BAs of 0.729, 0.908, and 0.980 across OAR types. Our novel knowledge-based method combines models utilizing single-contour and CCR features to effectively detect erroneous OAR contours across a comprehensive set of 42 clinically used OAR types for HN radiotherapy.

LSTM Auto Encoder 이용한 접근 이상 항적 탐지 모형

With the rise in aviation demand and the emergence of Urban Air Mobility, developing a safe aviation system in urban areas is becoming increasingly important. This study addresses the challenge of detecting anomalous flight trajectories, which can be influenced by environmental factors. We propose a novel Long Short-Term Memory-Auto Encoder (LSTM-AE) model that processes both environmental and trajectory data but only reconstructs trajectory data in its output. This approach was validated by assessing the average reconstruction error for specific trajectories. Additionally, the model's ability to identify anomalies was confirmed by evaluating the Area under the ROC curve (AUC) for typical anomalous trajectories, such as go-around maneuvers. Our findings indicate that the proposed LSTM-AE model effectively learns trajectory patterns in relation to environmental variables and shows enhanced anomaly detection capabilities compared to traditional AE and LSTM-AE models. These results contribute to the development of advanced models that incorporate a wider range of environmental factors, enhancing safety in urban air travel.

Transformer-Based Representation of Organic Molecules for Potential Modeling of Physicochemical Properties.

In this work, we study the use of three configurations of an autoencoder neural network to process organic substances with the aim of generating meaningful molecular descriptors that can be employed to develop property prediction models. A total of 18,322,500 compounds represented as SMILES strings were used to train the model, demonstrating that a latent space of 24 units is able to adequately reconstruct the data. After AE training, an analysis of the latent space properties in terms of compound similarity was carried out, indicating that this space possesses desired properties for the potential development of models for forecasting physical properties of organic compounds. As a final step, a QSPR model was developed to predict the boiling point of chemical substances based on the AE descriptors. 5276 substances were used for the regression task, and the predictive ability was compared with models available in the literature evaluated on the same database. The final AE model has an overall error of 1.40% (1.39% with augmented SMILES) in the prediction of the boiling temperature, while other models have errors between 2.0 and 3.2%. This shows that the SMILES representation is comparable and even outperforms the state-of-the-art representations widely used in the literature.

Voxel representation of brain images inpainting via Regional Pixel Semantic Network and pyramidal attention AE - Quantile differential mechanism model

Medical image inpainting holds significant importance in enhancing the quality of medical images by restoring missing areas, thereby rendering them suitable for diagnostic purposes. While several techniques have been previously proposed for medical image inpainting, they are not suitable for distorted images containing metallic implants due to their limited consideration of known shaped masking. To overcome this limitation, a novel Vectorized Box Interpolation with Arbitrary Auto-Rand Augment Masking technique has been proposed which involves scaling and vectorizing images to expand their details and generating asymmetrically shaped masking in an automatic random format. One of the challenging tasks in this regard is the precise detection of lost regions, which is addressed through the introduction of the Regional Pixel Semantic Network. This technique employs the locally shared features (LSF) based region sensing with FCN (fully convolutional network) segmentation, which performs automatic segmentation based on neighboring pixel local dependency and regional features to determine the location of masked regions. During the reconstruction of missing parts, a significant challenge posed is the inability to recognize proximity in encoding owing to the generation of shadow-like regions on the feature map. To address this issue, a novel Multilayered DRC Regularized Pyramidal Attention AE Model has been proposed which employs dilated convolution with coherent pyramidal attention for feature extraction and improves image resolution using a Laplacian convolutional layer. Moreover, the realness of the generated image is determined using the Quantile Differential Mechanism model, where in the Quantile Differential Partial Convolutional Discriminator utilizes the hyperbolic tangent activation function in the partial convolutional layer to calculate recognition accuracy. As a result, the proposed method achieves high percentages for accuracy (98 %), precision (97 %), sensitivity (96 %), recall (95 %), and F-measure (96 %) thereby outperforming existing methods. Overall, this proposed method effectively handles distorted images with metallic implants, accurately detects lost regions, and improves the reconstructed image quality.

GAUSS: Guided encoder - decoder Architecture for hyperspectral Unmixing with Spatial Smoothness

ABSTRACT This study introduces GAUSS (Guided encoder-decoder Architecture for hyperspectral Unmixing with Spatial Smoothness), a novel autoencoder-based architecture for hyperspectral unmixing (HU). GAUSS consists of an Approximation Network (AN), Unmixing Network (UN), and a Mixing Network (MN). The AN incorporates spatial context within a hyperspectral pixel’s neighborhood, while the UN utilizes a pseudo-ground truth mechanism to enhance abundance estimation. The MN provides estimated endmembers’ signatures. By incorporating UN-produced abundances, unlike the conventional AE model, GAUSS overcomes the single-layer constraint of the MN. Thereafter, a secondary training phase improves the accuracy of endmembers and abundance estimation using a reliable Signal Processing (SP) algorithm, resulting in superior HU performance. The results demonstrate the effectiveness of GAUSS on two Standard datasets and a Simulated dataset compared to the state-of-the-art SP and Deep Learning (DL) based methods. This signifies the benefit of integrating an SP algorithm in the training process, contributing to advancements in DL-based HU techniques.

A novel autoencoder modeling method for intelligent assessment of bearing health based on Short-Time Fourier Transform and ensemble strategy

In recent years, deep learning (DL) has developed rapidly in mechanical fault diagnosis. However, in the process of developing rolling bearing fault diagnosis DL model, especially for multiple-category fault diagnosis, the selection of hyperparameters of DL models is critical and difficult. This paper proposed a novel method for an intelligent assessment of bearing health based on Short-Time Fourier Transform (STFT) and ensemble strategy, namely Ensemble STFT-multi-AEs networks (ESAEs). The method was composed of four steps. Firstly, the vibration signal was converted into three-channel time-frequency maps by the STFT. Secondly, three multi-autoencoders (multi-AEs) networks were established using the three-channel time-frequency maps, and the multi-channel features extracted by the three networks were fused. Thirdly, a series of multi-AE networks were built based on 11 activation functions (AFs) with different characteristics such as output ranges, monotonicity, smoothness, etc. The consensus strategy was adopted to integrate the modeling results of each sub-model, and the advantages of each member model were analyzed. Finally, a pyramid grading strategy was applied to the member models with larger weights in the 11 sub-models for the second allocation of weights to eliminate the influence of extreme models on the stability of ensemble results. The results show that the ESAEs method can effectively and automatically extract the deep vibration features of bearings, reducing the parameter selection steps brought by AFs. The pyramid grading strategy can effectively balance the difference of prediction results of AE models built with different AFs. The ESAEs model achieved the best detection results for locomotive bearing defects with an accuracy of 99.2%. In addition, the ESAEs method still maintains good detection performance and stability in the fault diagnosis of data imbalance.