Deep Convolutional Autoencoder Research Articles

New Health Indicator Construction and Fault Detection Network for Rolling Bearings via Convolutional Auto-Encoder and Contrast Learning

As one of the most important components in rotating machinery, if bearings fail, serious disasters may occur. Therefore, the remaining useful life (RUL) prediction of bearings is of great significance. Health indicator (HI) construction and early fault detection play a crucial role in data-driven RUL prediction. Unfortunately, most existing HI construction methods require prior knowledge and preset trends, making it difficult to reflect the actual degradation trend of bearings. And the existing early fault detection methods rely on massive historical data, yet manual annotation is time-consuming and laborious. To address the above issues, a novel deep convolutional auto-encoder (CAE) based on envelope spectral feature extraction is developed in this work. A sliding value window is defined in the envelope spectrum to obtain initial health indicators, which are used as preliminary labels for model training. Subsequently, CAE is trained by minimizing the composite loss function. The proposed construction method can reflect the actual degradation trend of bearings. Afterwards, the autoencoder is pre-trained through contrast learning (CL) to improve its discriminative ability. The model that has undergone offline pre-training is more sensitive to early faults. Finally, the HI construction method is combined with the early fault detection method to obtain a comprehensive network for online health assessment and fault detection, thus laying a solid foundation for subsequent RUL prediction. The superiority of the proposed method has been verified through experiments.

Single-Pixel Imaging Based on Deep Learning Enhanced Singular Value Decomposition.

We propose and demonstrate a single-pixel imaging method based on deep learning network enhanced singular value decomposition. The theoretical framework and the experimental implementation are elaborated and compared with the conventional methods based on Hadamard patterns or deep convolutional autoencoder network. Simulation and experimental results show that the proposed approach is capable of reconstructing images with better quality especially under a low sampling ratio down to 3.12%, or with fewer measurements or shorter acquisition time if the image quality is given. We further demonstrate that it has better anti-noise performance by introducing noises in the SPI systems, and we show that it has better generalizability by applying the systems to targets outside the training dataset. We expect that the developed method will find potential applications based on single-pixel imaging beyond the visible regime.

Prediction Model of Ancient Village Pottery Building Microspace Design Style by Integrating Machine Learning

Microspace design styles on pottery fragments offer a captivating window into the artistic heritage and cultural practices of past societies. These intricate details and recurring patterns hold valuable clues about rituals, beliefs, and artistic expressions. However, existing approaches struggle to capture the relationships between design elements within a single piece. This limitation hinders the ability to capture the holistic meaning conveyed by the microspace design style. To overcome this limitation, this research proposes a novel approach called MoANN-DSOA for predicting microspace design styles based on ancient village pottery. MoANN-DSOA utilizes a Mosaic Attention Neural Network (MoANN) to analyze both the visual image and the relationships between design elements. This allows for a more detailed understanding of the artistic message encoded within each fragment. Additionally, a Dove Swarm Optimization Algorithm (DSOA) optimizes the MoANN architecture, potentially enhancing the accuracy of capturing intricate details. The proposed MoANN-DSOA method attains 7.78%, 27.89% and 33.335% higher accuracy and 14%, 20.81% and 32.36% higher F-Score compared to the existing methods like Categorization and Retrieval of Painted Pottery with CNNs (CNN), Utilizing CNN-VGG16-VGG19 Approach for Distinguishing Surface Treatments in Wheel-Thrown Pottery (CNN-VGG16-VGG19) and Unsupervised Feature Extraction of Ceramic Profiles using a Deep Variational Convolutional Autoencoder (DVCA) respectively. By this, the proposed MoANN-DSOA methodology paves the way for efficient information extraction from pottery image, unlocking deeper understanding and facilitating the construction of more informed historical narratives.

Modelling appearance variations in expressive and neutral face image for automatic facial expression recognition

AbstractIn automatic facial expression recognition (AFER) systems, modelling the spatio‐temporal feature information in a specific manner, coalescing, and its effective utilization is challenging. The state‐of‐the‐art studies have examined integrating multiple features to enhance the recognition rate of AFER systems. However, the feature variations between expressive and neutral face images are not fully explored to identify the expression class. The proposed research presents an innovative approach to AFER by modelling appearance variations in both expressive and neutral face images. The prominent contributions of the work are developing a novel and hybrid feature space by integrating the discriminative feature distribution derived from expressive and neutral face images; preserving the highly discriminative latent feature distribution using autoencoders. Local binary pattern (LBP) and histogram of oriented gradients (HOG) are the feature descriptors employed to derive the discriminative texture and shape information, respectively. The component‐based approach is employed, wherein the features are derived from the salient facial regions instead of the whole face. The three‐stage stacked deep convolutional autoencoder (SDCA) and multi‐class support vector machine (MSVM) are employed to address dimensionality reduction and classification, respectively. The efficacy of the proposed model is substantiated by empirical findings, which establish its superiority in terms of accuracy in AFER tasks on widely recognized benchmark datasets.

Open Set Bearing Fault Diagnosis with Domain Adaptive Adversarial Network under Varying Conditions

Bearing fault diagnosis is a pivotal aspect of monitoring rotating machinery. Recently, numerous deep learning models have been developed for intelligent bearing fault diagnosis. However, these models have typically been established based on two key assumptions: (1) that identical fault categories exist in both the training and testing datasets, and (2) the datasets used for testing and training are assumed to follow the same distribution. Nevertheless, these assumptions prove impractical and fail to accurately depict real-world scenarios, particularly those involving open-world assumption fault diagnosis in multi-condition scenarios. For that purpose, an open set domain adaptive adversarial network framework is proposed. Specifically, in order to improve the learning of distribution characteristics in different fields, comprehensive training is implemented using a deep convolutional autoencoder model. Additionally, to mitigate the negative transfer resulting from unknown fault samples in the target domain, the similarity of each target domain sample and the shared classes in the source domain are estimated using known class classifiers and extended classifiers. Similarity weight values are assigned to each target domain sample, and an unknown boundary is established in a weighted manner. This approach is employed to establish the alignment between the classes shared between the two domains, enabling the classification of known fault classes, while allowing the recognition of unknown fault classes in the target domain. The efficacy of our suggested approach is empirically validated using different datasets.

Hierarchical functional differences between gyri and sulci at different scales.

Gyri and sulci are 2 fundamental cortical folding patterns of the human brain. Recent studies have suggested that gyri and sulci may play different functional roles given their structural and functional heterogeneity. However, our understanding of the functional differences between gyri and sulci remains limited due to several factors. Firstly, previous studies have typically focused on either the spatial or temporal domain, neglecting the inherently spatiotemporal nature of brain functions. Secondly, analyses have often been restricted to either local or global scales, leaving the question of hierarchical functional differences unresolved. Lastly, there has been a lack of appropriate analytical tools for interpreting the hierarchical spatiotemporal features that could provide insights into these differences. To overcome these limitations, in this paper, we proposed a novel hierarchical interpretable autoencoder (HIAE) to explore the hierarchical functional difference between gyri and sulci. Central to our approach is its capability to extract hierarchical features via a deep convolutional autoencoder and then to map these features into an embedding vector using a carefully designed feature interpreter. This process transforms the features into interpretable spatiotemporal patterns, which are pivotal in investigating the functional disparities between gyri and sulci. We evaluate the proposed framework on Human Connectome Project task functional magnetic resonance imaging dataset. The experiments demonstrate that the HIAE model can effectively extract and interpret hierarchical spatiotemporal features that are neuroscientifically meaningful. The analyses based on the interpreted features suggest that gyri are more globally activated, whereas sulci are more locally activated, demonstrating a distinct transition in activation patterns as the scale shifts from local to global. Overall, our study provides novel insights into the brain's anatomy-function relationship.

Identification of multi-mineral-species geochemical anomalies using Bayesian maximum entropy and the spectrum separable module-constrained convolutional autoencoder

In this paper, we present a dual-drive multi-mineral-species anomaly detection system which involves the combined use of Bayesian maximum entropy, a spectral separable module, high-order factor analysis, a geologically constrained loss function, as well as a mixed Gaussian distribution-based thresholding algorithm, attaching to a deep convolutional autoencoder. We have achieved several firsts rarely considered in the existing literature, for example: previous works focus mainly on separating single-mineral-species rather than multi-elemental anomalies, while we have attempted to recognize multi-mineral-species anomalies; previous works pay more attention to data, while we suggest how to discover the ore-related correlations hidden within the input data; previous works fail to integrate soft data in a quantitative fashion, while we have achieved that by capitalizing on Bayesian maximum entropy and the stratigraphic combination entropy. A series of comparative experiments have demonstrated the advantages over other state-of-the-art approaches. Finally, we obtained a mineral-occurrence identification rate ( δ ) ranging from 36.87 to 61.46% v. the anomaly area ranging from 33.75 to 55.38% for each metalliferous anomaly division.

One-dimensional deep convolutional autoencoder active infrared thermography: Enhanced visualization of internal defects in FRP composites

Fiber-reinforced polymer (FRP) composites have been widely applied in different industrial fields, thereby necessitating the employment of non-destructive testing (NDT) methods to ensure structural integrity and safety. Active infrared thermography (AIRT) is a fast and cost-efficient NDT technique for inspecting FRP composites. However, this method is easily affected by factors such as inhomogeneous heating, leading to a low level of visualization of defects. To address this issue, this study proposes a novel method called one-dimensional deep convolutional autoencoder active infrared thermography (1D-DCAE-AIRT) to enhance the visualization of internal defects in FRP composites. This method first preprocesses the thermal image sequence acquired by AIRT inspections. Subsequently, high-level thermal features at the pixel level are extracted from the aforementioned preprocessed thermal image sequence using a designed one-dimensional deep convolutional autoencoder (1D-DCAE) model. Finally, the extracted high-level thermal features are employed to generate enhanced visualization results that exhibit improved defect visibility. The results of three kinds of AIRT (eddy current pulsed thermography, flash thermography, and vibrothermography) experiments on FRP composite specimens with artificially introduced defects show that 1D-DCAE-AIRT can effectively enhance the visualization of internal defects. The enhancement effect is better than the conventional techniques of fast Fourier transform (FFT), principal component analysis (PCA), independent component analysis (ICA), and partial least-squares regression (PLSR).

HRTF low-dimensional representation based on deep convolutional autoencoder and attention mechanism

HRTF low-dimensional representation based on deep convolutional autoencoder and attention mechanism

The integration of machine learning and IoT for the early detection of tomato leaf disease in real-time

The impact of climate change, pests, and inadequate agricultural practices on crop health is becoming a growing concern. It is estimated that 20-40% of global crop yield is adversely affected by pests and diseases. This has a direct negative impact on food security and nutritional well-being, as staple cereal crops (such as rice, maize, and wheat) and tuber crops (like potato, onion, and tomato) are affected. Advancements in Artificial Intelligence (AI), Computer Vision (CV), and IoT have a significant influence on reducing crop losses by detecting crop diseases at early stages. The primary focus of this research is to develop a real-time system that can detect tomato crop diseases at an early stage by integrating Machine Learning (ML) algorithms and IoT. The most common tomato leaf diseases, such as Tomato Mosaic Virus (TMV), Tomato Bacterial Leaf Spot (TBLS), Tomato Early Blight (TEB), and Tomato Late Blight (TLB), are considered in this work. The hybrid discriminative feature space is derived from the integration of low- and high-level features via Convolution Neural Network (CNN) Layers. The 3-stage Stacked Deep Convolutional Autoencoder is used to optimize the CNN performance by reducing computation complexity. The proposed model is implemented on the Plantvillage benchmark dataset and achieves the highest recognition accuracy of 95.6% for the 5-class problem using 5-fold cross-validation.

Feature Extraction in Time-Lapse Seismic Using Deep Learning for Data Assimilation

Summary Assimilation of time-lapse (4D) seismic data with ensemble-based methods is challenging because of the massive number of data points. This situation requires excessive computational time and memory usage during the model updating step. We addressed this problem using a deep convolutional autoencoder to extract the relevant features of 4D images and generate a reduced representation of the data. The architecture of the autoencoder is based on the VGG-19 network, a deep convolutional architecture with 19 layers well-known for its effectiveness in image classification and object recognition. Some advantages of VGG-19 are the possibility of using some pretrained convolutional layers to create a feature extractor and taking advantage of the transfer learning technique to address other related problem domains. Using a pretrained model bypasses the need for large training data sets and avoids the high computational demand to train a deep network. For further improvements in the reconstruction of the seismic images, we apply a fine-tuning of the weights of the latent convolutional layer. We propose to use a fully convolutional architecture, which allows the application of distance-based localization during data assimilation with the ensemble smoother with multiple data assimilation (ES-MDA). The performance of the proposed method is investigated in a synthetic benchmark problem with realistic settings. We evaluate the methodology with three variants of the autoencoder, each one with a different level of data reduction. The experiments indicate that it is possible to use latent representations with major data reductions without impairing the quality of the data assimilation. Additionally, we compare central processing unit (CPU) and graphics processing unit (GPU) implementations of the ES-MDA update step and show in another synthetic problem that the reduction in the number of data points obtained with the application of the deep autoencoder may provide a substantial improvement in the overall computation cost of the data assimilation for large reservoir models.

A novel approach for sports injury risk prediction: based on time-series image encoding and deep learning.

The rapid development of big data technology and artificial intelligence has provided a new perspective on sports injury prevention. Although data-driven algorithms have achieved some valuable results in the field of sports injury risk assessment, the lack of sufficient generalization of models and the inability to automate feature extraction have made it challenging to deploy research results in the real world. Therefore, this study attempts to build an injury risk prediction model using a combination of time-series image encoding and deep learning algorithms to address this issue better. This study used the time-series image encoding approach for feature construction to represent relationships between values at different moments, including Gramian Angular Summation Field (GASF), Gramian Angular Difference Field (GADF), Markov Transition Field (MTF), and Recurrence Plot (RP). Deep Convolutional Auto-Encoder (DCAE) learned the image-encoded data for representation to obtain features with good discrimination, and the classifier was performed using Deep Neural Network (DNN). The results from five repeated experiments show that the GASF-DCAE-DNN model is overall better in the training (AUC: 0.985 ± 0.001, Gmean: 0.930 ± 0.007, Sensitivity: 0.997 ± 0.003, Specificity: 0.868 ± 0.013) and test sets (AUC: 0.891 ± 0.026, Gmean: 0.830 ± 0.027, Sensitivity: 0.816 ± 0.039, Specificity: 0.845 ± 0.022), with good discriminative power, robustness, and generalization ability. Compared with the best model reported in the literature, the AUC, Gmean, Sensitivity, and Specificity of the GASF-DCAE-DNN model were higher by 23.9%, 27.5%, 39.7%, and 16.2%, respectively, which confirmed the validity and practicability of the model in injury risk prediction. In addition, differences in injury risk patterns between the training and test sets were identified through shapley additivity interpretation. It was also found that the training volume was an essential factor that affected injury risk prediction. The model proposed in this study provides a powerful injury risk prediction tool for future sports injury prevention practice.

DANTD: A Deep Abnormal Network Traffic Detection Model for Security of Industrial Internet of Things Using High-Order Features

With the development of blockchain, artificial intelligence and data mining technology, abnormal network traffic data has become easy to obtain. The traffic detection model detects the traffic patterns in the network to find abnormal traffic that does not conform to the normal traffic law, which has great security significance for Industrial Internet of Things (IIoT) networks and devices in real scenarios. However, previous abnormal detection models rely on expert experience and cannot cope with real-time changes in IIoT scenarios. The manual features cannot be sufficiently representative and adaptive. Moreover, there are few abnormal traffic data in real scenarios, which makes the model unable to fully learn the potential distribution in abnormal data. Therefore, in this work, we propose a deep abnormal network traffic detection model (DANTD) for security of IIoT using high-order features and novel data augmentation strategies. The DANTD model first adopts a deep convolutional autoencoder to extract effective high-order features to make it more representative. Then the DANTD model uses generative adversarial networks as data augmentation strategies to enrich the abnormal data, so that the model can fully consider the information of the data distribution. Comprehensive experiments on real IIoT datasets validate the effectiveness of the DANTD model.

Road crack detection interpreting background images by convolutional neural networks and a self‐organizing map

AbstractThe presence of road cracks is an important indicator of damage. Deep learning is a prevailing method for detecting cracks in road surface images because of its detection ability. Previous research works focused on supervised convolutional neural networks (CNNs) without non‐crack features or unsupervised crack analysis with limited accuracies. The novelty of this study is the addition of background classification. By increasing the number of non‐crack categories, CNNs are driven to learn non‐crack features and improve crack detection performances. Non‐crack images are preprocessed, and their features are extracted in an unsupervised way by a deep convolutional autoencoder. A self‐organizing map clusters features to obtain non‐crack categories. This study focusses on classification though the method can be adopted in parallel with the latest segmentation algorithms. Using common road crack datasets, modified deep CNN models significantly improved accuracy by 1%–4% and f‐measure by 3%–8%, compared to previous models. The modified visual geometry group (VGG) 16 showed the top‐level performance, 96% accuracy and 84%–85% f‐measure. The models drastically reduced false detection cases while maintaining their crack detection abilities.

A Novel Approach to Flexible Multi-Resolution Image Compression using Deep Learning Based Autoencoders on Overlapping Image Patch

Compressing images using a deep-learning-based autoencoder usually can only work for images of a fixed size. This causes problems since the images might be in various sizes. This paper proposes a novel approach that can be applied in real-life based on a patching algorithm for a given image. This algorithm aims to crop the image into smaller ones that can fit into our autoencoder model so that a single autoencoder model can be used for images with diverse sizes. The proposed approach was tested on a few autoencoder types (Vanilla, Deep Convolutional, and Variational Autoencoder), and it was found that different types of autoencoder impact the quality of the recreated images based on PSNR (Peak signal-to-noise ratio), MSE (Mean squared error), and SSIM (Structural Similarities). In our research we found that for PSNR and MSE, Variational Autoencoder got highest score among other design, with average of 21 PSNR and 1300 MSE Score and Deep Convolutional Autoencoder gives the best results in SSIM score among all other types of autoencoders with SSIM Score rate average at 0.5, when implemented in our system.

Morphological diversity of cancer cells predicts prognosis across tumor types.

Intratumor heterogeneity drives disease progression and treatment resistance, which can lead to poor patient outcomes. Here, we present a computational approach for quantification of cancer cell diversity in routine hematoxylin-eosin-stained histopathology images. We analyzed publicly available digitized whole-slide hematoxylin-eosin images for 2000 patients. Four tumor types were included: lung, head and neck, colon, and rectal cancers, representing major histology subtypes (adenocarcinomas and squamous cell carcinomas). We performed single-cell analysis on hematoxylin-eosin images and trained a deep convolutional autoencoder to automatically learn feature representations of individual cancer nuclei. We then computed features of intranuclear variability and internuclear diversity to quantify tumor heterogeneity. Finally, we used these features to build a machine-learning model to predict patient prognosis. A total of 68 million cancer cells were segmented and analyzed for nuclear image features. We discovered multiple morphological subtypes of cancer cells (range = 15-20) that co-exist within the same tumor, each with distinct phenotypic characteristics. Moreover, we showed that a higher morphological diversity is associated with chromosome instability and genomic aneuploidy. A machine-learning model based on morphological diversity demonstrated independent prognostic values across tumor types (hazard ratio range = 1.62-3.23, P < .035) in validation cohorts and further improved prognostication when combined with clinical risk factors. Our study provides a practical approach for quantifying intratumor heterogeneity based on routine histopathology images. The cancer cell diversity score can be used to refine risk stratification and inform personalized treatment strategies.

Deep semi-supervised electricity theft detection in AMI for sustainable and secure smart grids

Electricity theft is a major issue that affects the sustainability and security of smart grids. This paper proposes a deep semi-supervised method for electricity theft detection in the Advanced Metering Infrastructure (AMI) of smart grids. By only using normal samples to train the detection model, the proposed method has the capability of detecting unknown attacks in a short time frame. The method utilizes the ratio profile generated from the readings of the observer meter and the user’s smart meter as the input, to reduce false positives, which is then transformed into a 2D image with the continuous wavelet transform (CWT) to capture the time–frequency information. Discriminative features are extracted from the CWT image with a deep convolutional autoencoder (CAE) and principal component analysis (PCA), which are fed into a semi-supervised autoencoder for classification. The performance of the proposed method was evaluated and compared with a set of baselines and four supervised machine learning and deep learning methods under 11 different false data injection (FDI) attacks using smart meter data from both business and residential users. The results show that the proposed method significantly outperforms the baselines and is more capable of detecting unknown attacks than supervised methods.

Fault state identification of rolling bearings based on deep transfer convolutional autoencoder and a new health indicator

For the problem of bearing fault state identification, in recent years, the main research is how to distinguish the normal state and the fault state. However, it is still difficult to accurately identify the more detailed fault severity in the cross-working environment. Therefore, this paper proposes a new health indicator (HI) and deep transferable convolutional autoencoder network (DTCAE) to solve the problem of rolling bearing early fault monitoring and fault state recognition. Firstly, the envelope spectrum of the vibration signal is calculated, and then the ratio of the energy at the maximum amplitude of the envelope spectrum to other energies is obtained, and the final HI is obtained by weighting the standard deviation with it, which has good generalization and engineering practicability. HI is used for early bearing fault monitoring and classifying bearing fault status in source domain. Finally, aiming at the problem that the existing transfer learning methods do not consider the weakening of the target domain data features after domain adaptation, DTCAE is proposed to identify the bearing fault state. Specifically, in addition to the distribution difference loss and classification loss commonly seen in transfer learning models, a decoder is also added after feature extraction to calculate the difference between the reconstructed data of the target domain and the original data, and the feature distribution difference between the reconstructed source domain and the target domain is also calculated. This allows the source-domain bearing and the target-domain bearing features to be matched in such a way that both not only have the same feature distribution, but also retain the degradation trend characteristics of the target-domain bearing. Experiments in three challenging cases have achieved good results and are superior to some existing methods.

Reciprocating compressor health monitoring based on BSInformer with deep convolutional AutoEncoder

A composite model combining deep convolutional autoencoder and balanced sparse sampling Informer (MCA-BSInformer) is proposed to solve the problem that reciprocating compressors are difficult to achieve long-term accurate health monitoring. By introducing the Maximum Mean Discrepancy (MMD) difference term into the deep convolutional autoencoder, the distribution of noise data in the multi-source signal of the compressor become more similar to the distribution of normal training data. Thus, effective low-dimensional features containing spatial information in the data can be obtained to reduce the impact of noise data. In addition, a balanced sparse sampling algorithm is integrated into the Informer model to achieve adaptive sparse sampling. The improved Informer model effectively solves the problem that the model reduces the prediction performance while pursuing efficiency. The experimental results demonstrate that the prediction accuracy of the composite model for the health status of reciprocating compressors is higher than that of other mainstream models.

A mixed‐attention‐based multi‐scale autoencoder algorithm for fabric defect detection

AbstractAiming at the defects in the process of fabric production, a defect detection model of fabric based on a mixed‐attention‐based multi‐scale non‐skipping U‐shaped deep convolutional autoencoder (MADCAE) was proposed. In a traditional encoder, the convolutional layer treats each pixel equally, so the importance of different pixels cannot be reflected. It is difficult to obtain richer and more effective information. The reconstruction of the defect region and the detection of the defect region are further affected. In this article, three different scale features of input images are extracted by enlarging the receptive field with large kernel convolution blocks. A hybrid attention module is used to ensure the richness of the extracted information in terms of space and channel. Experiments show that this method can effectively reconstruct fabric parts without requiring a large number of defect marking samples. It can quickly detect and locate defective areas of fabric patterns.