Kernel methods and neural networks (NNs) are two mainstream nonlinear data modeling methods and have been widely applied to industrial process monitoring. However, they both present imperfect properties, so the releva...
详细信息
Kernel methods and neural networks (NNs) are two mainstream nonlinear data modeling methods and have been widely applied to industrial process monitoring. However, they both present imperfect properties, so the relevant applications are limited. On the one hand, kernels are not so reconstructable, scalable, and robust to hyperparameters that they suffer performance degradation for large-scale data modeling and monitoring. On the other hand, the high-dimensional parameter space of NNs that is sorted to parameter initialization presents severe anomaly detection performance inconsistency, which makes the industry cautious about using NNs. Motivated by these facts, we propose to integrate kernels and NNs, forming a new model structure that is scalable, reconstructable, and performance-consistent. Specifically, a novel autoencoder-based nonstationary pattern selection kernel (AE-NPSK) is proposed by (1) selecting from the training set the critical edges and interior data as the centers of the radial basis functions in the hidden layers and (2) adaptively adjusting the kernel width in the training procedure. Also, the new NN has strong performance consistency, which facilitates the search for optimal parameters. Finally, we test the performance of the proposed method on the challenging multimode processes. The results validate the efficacy of the proposed method.
作者：
Han, MingDang, YuHan, JiandaNankai Univ
Inst Robot & Automatic Informat Syst Coll Artificial Intelligence Tianjin 300350 Peoples R China Nankai Univ
Engn Res Ctr Trusted Behav Intelligence Minist Educ Tianjin 300350 Peoples R China Nankai Univ
Tianjin Key Lab Intelligent Robot Tianjin 300350 Peoples R China Nankai Univ
Inst Intelligence Technol & Robot Syst Shenzhen Res Inst Shenzhen 518083 Peoples R China
Preprocessing plays a key role in Raman spectral analysis. However, classical preprocessing algorithms often have issues with reducing Raman peak intensities and changing the peak shape when processing spectra. This p...
详细信息
Preprocessing plays a key role in Raman spectral analysis. However, classical preprocessing algorithms often have issues with reducing Raman peak intensities and changing the peak shape when processing spectra. This paper introduces a unified solution for preprocessing based on a convolutional autoencoder to enhance Raman spectroscopy data. One is a denoising algorithm that uses a convolutional denoising autoencoder (CDAE model), and the other is a baseline correction algorithm based on a convolutional autoencoder (CAE+ model). The CDAE model incorporates two additional convolutional layers in its bottleneck layer for enhanced noise reduction. The CAE+ model not only adds convolutional layers at the bottleneck but also includes a comparison function after the decoding for effective baseline correction. The proposed models were validated using both simulated spectra and experimental spectra measured with a Raman spectrometer system. Comparing their performance with that of traditional signal processing techniques, the results of the CDAE-CAE+ model show improvements in noise reduction and Raman peak preservation.
In recent years, the increased application of controller area network (CAN) protocols has made it the de facto standard for communication between electronic control units (ECUs) in the automotive and transportation fi...
详细信息
In recent years, the increased application of controller area network (CAN) protocols has made it the de facto standard for communication between electronic control units (ECUs) in the automotive and transportation fields. This widely used protocol was designed as a reliable and straightforward broadcast-based protocol that connects ECUs without considering security concerns such as node authentication or traffic encryption. Despite its efficiency, this tradeoff makes the CAN bus vulnerable to attacks. Implementing intrusion detection systems (IDSs) based on machine learning (ML) can address these security challenges effectively. However, existing ML-based IDSs have limited classification capabilities, lack adaptability and time sensitivity, incomprehensive analysis, and produce high false-negative rates (FNR), while attack schemes are becoming increasingly complex, resulting in insufficient capability of intrusion detection in real-time and insufficient ability to offer reliable protection. Therefore, our study proposes a novel in-vehicle IDS for multiclass classification using both packet- and sequence-level characteristics extracted from an autoencoder and a variant transformer (Time-embedded Transformer) with an improved position encoding mechanism, which analyses CAN traffic in-depth from various perspectives to overcome the existing challenges above. Both standard (Car-Hacking) and advanced (ROAD) datasets are used to evaluate the capabilities of our proposed IDS. The evaluation results demonstrated 100 % detection accuracy and 0 % FNR for both the Car-Hacking and ROAD Masquerade datasets, which also peaked at the highest F1 score for the ROAD Fabrication dataset, emphasizing superior intrusion detection to minimize FNR of the proposed model with high adaptability through its multi-dimensional analysis at packet- and sequencelevel.
This paper proposes refrigerated showcase fault detection by an autoencoder with the adaptive kernel size tuning (AKST) using Maximum Correntropy Criterion (MCC) and Stochastic Gradient Descent with Momentum. Refriger...
详细信息
ISBN:
(纸本)9781665487689
This paper proposes refrigerated showcase fault detection by an autoencoder with the adaptive kernel size tuning (AKST) using Maximum Correntropy Criterion (MCC) and Stochastic Gradient Descent with Momentum. Refrigerated showcase data may include outliers that store values differing from actual values because of various reasons such as incorrect sensor settings and radio frequency interference. The conventional fault detection methods by an autoencoder using Least Square Error (LSE) as a loss function are affected by the outliers when the outliers are included in learning data. On the contrary, the conventional fault detection methods by an Artificial Neural Network (ANN) using MCC as a loss function can ignore influence of the outliers. Moreover, faults rarely occur in refrigerated showcases. The methods by an ANN using MCC require fault data for learning. A fault detection method by an autoencoder using the MCC that is a combination of the above two methods can ignore influence of the outliers and doesn't require fault data. However, there is a parameter named kernel size in the MCC. It is required to tune the parameter properly and the parameter tuning requires engineering. The proposed AKST method reduces the engineering of the kernel size tuning. Effectiveness of the proposed method is confirmed by comparison with comparative methods by an autoencoder using LSE and Stochastic Gradient Descent (SGD), an autoencoder with fixed kernel size using the MCC and SGD, an autoencoder with the AKST using the MCC and Adam, and an autoencoder with the AKST using the MCC and AdaGrad.
Semi-supervised detection of outliers with only positive and unlabeled data, which is among the most frequent forms of the anomaly detection (AD) problem in real scenarios, requires for a model to capture the normal b...
详细信息
Semi-supervised detection of outliers with only positive and unlabeled data, which is among the most frequent forms of the anomaly detection (AD) problem in real scenarios, requires for a model to capture the normal behaviour of data from a training set exclusively comprised of normal-labelled data, so new unseen data can be afterwards compared to the induced notion of normality to be flagged -or not- as anomalous. In modelling a certain pattern of behaviour, generative models such as generative-adversarial networks (GANs) have proved to have great performance. Thus, numerous AD algorithms with GANs at its core have been proposed, most of them powered by deep neural networks and relying on an autoencoder for the AD task. In the present work, a novel approach to semi-supervised AD with Bayesian networks using generative-adversarial training and an evolutive strategy is proposed, which aims to palliate the intrinsic lack of interpretability of deep neural networks. The proposed model is tested on a real-world AD problem in cybersecurity.
Recently, the communication system has evolved from 5G to 6G, and the integration of communication sensing is a major trend. In order to jointly optimize the communication and sensing performance and eliminate the int...
详细信息
ISBN:
(纸本)9781665469012
Recently, the communication system has evolved from 5G to 6G, and the integration of communication sensing is a major trend. In order to jointly optimize the communication and sensing performance and eliminate the interference in the transmission process, this paper proposes a signal interference cancellation scheme based on autoencoder to improve the signal quality of the communication signal and the sensing signal. The proposed algorithm model can effectively unify the transmission data formats of different structural forms, and perform joint reconstruction in a general way. In particular, an Inception model is added to the decoder section to improve performance. Based on the simulation software, the transmission system of the OFDM signal format is built, and the effectiveness of the proposed scheme has been confirmed.
Hyperspectral underwater target detection (HUTD) is a promising and challenging task in remote sensing image processing. Existing methods face significant challenges when adapting to nearshore environments, where clut...
详细信息
Hyperspectral underwater target detection (HUTD) is a promising and challenging task in remote sensing image processing. Existing methods face significant challenges when adapting to nearshore environments, where cluttered backgrounds hinder the extraction of target signatures and exacerbate signal distortion. Hyperspectral unmixing (HU) demonstrates potential effectiveness for nearshore underwater target detection (UTD) by simultaneously extracting water background endmembers and separating target signals. To this end, this article investigates a novel nonlinear unmixing network for hyperspectral UTD, denoted as nonlinear unmixing network for hyperspectral-UTD (NUN-UTD), in which a well-designed autoencoder-based unmixing network is used to obtain the abundance map as the detection result. To address the weak underwater target signals, a target prior spectral preservation scheme is employed to guide the unmixing network in learning the accurate target abundance. Besides, to address the complexity of the nearshore environment, a pseudomixed data classification constraint is incorporated into the objective function to enhance the discriminative capability between the background and the target. Moreover, we adopt an additive postnonlinear model in the decoder to deal with the interactions between underwater spectra to account for the nonlinear effects between spectra of underwater substances. To validate the effectiveness of the proposed method, we constructed a hyperspectral dataset for nearshore UTD. Extensive experiments conducted on three real-world datasets and one simulated dataset demonstrate that our method achieves outstanding performance in HUTD.
The industrial application of data-driven methods for fault detection of new-design systems is limited by the inevitable scarcity of real data. Physics-Informed Neural Networks (PINNs) can mitigate this problem by int...
详细信息
The industrial application of data-driven methods for fault detection of new-design systems is limited by the inevitable scarcity of real data. Physics-Informed Neural Networks (PINNs) can mitigate this problem by integrating data and physical knowledge. In this work, we develop a novel fault detection method that combines physics-based simulations for data generation with a Physics-Informed Deep autoencoder (PIDAE) for reproducing the system behaviour in normal conditions;the Sequential Probability Ratio Test (SPRT) is, then, used for detecting abnormal conditions. The proposed method is applied to new-design electro-hydraulic servo actuators used in turbofan engine fuel systems. The results show that it can provide more satisfactory fault detection performance, in terms of false and missed alarms, than state-of-the-art methods based on traditional autoencoders only and pure physics-based models only. Furthermore, the PIDAE outcomes are physically consistent and, therefore, more acceptable and trustworthy.
Clustering plays a crucial role in the field of data mining, where deep non-negative matrix factorization (NMF) has attracted significant attention due to its effective data representation. However, deep matrix factor...
详细信息
Clustering plays a crucial role in the field of data mining, where deep non-negative matrix factorization (NMF) has attracted significant attention due to its effective data representation. However, deep matrix factorization based on autoencoder is typically constructed using multi-layer matrix factorization, which ignores nonlinear mapping and lacks learning rate to guide the update. To address these issues, this paper proposes an autoencoder-like deep NMF representation learning (ADNRL) algorithm for clustering. First, according to the principle of autoencoder, construct the objective function based on NMF. Then, decouple the elements in the matrix and apply the nonlinear activation function to enforce non-negative constraints on the elements. Subsequently, the gradient values corresponding to the elements update guided by the learning rate are transformed into the weight values. This weight values are combined with the activation function to construct the ADNRL deep network, and the objective function is minimized through the learning of the network. Finally, extensive experiments are conducted on eight datasets, and the results demonstrate the superior performance of ADNRL.
Representation learning is a crucial and challenging task within multimodal sentiment analysis. Effective multimodal sentiment representations contain two key aspects: consistency and difference. However, the state-of...
详细信息
Representation learning is a crucial and challenging task within multimodal sentiment analysis. Effective multimodal sentiment representations contain two key aspects: consistency and difference. However, the state-of-the-art multimodal sentiment analysis approaches failed to capture the difference and consistency of sentiment information across diverse modalities. To address the multimodal sentiment representation problem, we propose an autoencoder-based self -supervised learning framework. In the pre -training stage, an autoencoder is designed for each modality, leveraging unlabeled data to learn richer sentiment representations for each modality through sample reconstruction and modality consistency detection tasks. In the fine-tuning stage, the pre -trained autoencoder is injected into MulT (AE -MT) and enhance the model's ability to extract deep sentiment information by incorporating a contrastive learning auxiliary task. Our experiments on the popular Chinese sentiment analysis benchmark (CH-SIMS v2.0) and English sentiment analysis benchmark (MOSEI) demonstrate significant gains over baseline models.
暂无评论