作者:
Yizhe WangDeqing WangLiqun FuSchool of Informatics
and Key Laboratory of Underwater Acoustic Communication and Marine Information Technology of Ministry of EducationXiamen UniversityXiamen 361005China
The long delay spreads and significant Doppler effects of underwater acoustic(UWA)channels make the design of the UWA communication system more *** this paper,we propose a learning-based end-to-end framework for UWA c...
详细信息
The long delay spreads and significant Doppler effects of underwater acoustic(UWA)channels make the design of the UWA communication system more *** this paper,we propose a learning-based end-to-end framework for UWA communications,leveraging a double feature extraction network(DFEN)for data *** DFEN consists of an attentionbased module and a mixer-based module for channel feature extraction and data feature extraction,*** the diverse nature of UWA channels,we propose a stack-network with a two-step training strategy to enhance *** avoiding the use of pilot information,the proposed network can learn data mapping that is robust to UWA *** results show that our proposed algorithm outperforms the baselines by at least 2 dB under bit error rate(BER)10^(−2)on the simulation channel,and surpasses the compared neural network by at least 5 dB under BER 5×10^(−2)on the experiment channels.
Solar irradiance prediction is an essential subject in renewable energy generation. Prediction enhances the planning and management of solar installations and provides several economic benefits to energy companies. So...
详细信息
Solar irradiance prediction is an essential subject in renewable energy generation. Prediction enhances the planning and management of solar installations and provides several economic benefits to energy companies. Solar irradiation, being highly volatile and unpredictable makes the forecasting task complex and difficult. To address the shortcomings of the traditional approaches, this research developed a hybrid resilient architecture for an enhanced solar irradiation forecast by employing a long short-term memory (LSTM) autoencoder, convolutional neural network (CNN), and the Bi-directional Long Short Term Memory (BiLSTM) model with grid search optimization. The suggested hybrid technique is comprised of two parts: feature encoding and dimensionality reduction using an LSTM autoencoder, followed by a regularized convolutional BiLSTM. The encoder is tasked with extracting the key features in order to deduce the input into a compact latent representation. The decoder network then predicts solar irradiance by analyzing the encoded representation's attributes. The experiments are conducted on three publicly available data sets collected from Desert Knowledge Australia Solar Centre (DKASC), National Solar Radiation Database (NSRDB), and Hawaii Space Exploration Analog and Simulation (HI-SEAS) Habitat. The analysis of univariate and multivariate-multi step ahead forecasting performed independently and it is compared with the conventional approaches. Several benchmark forecasting models and three performance metrics are utilized to validate the hybrid approach's prediction performance. The results show that the proposed architecture outperforms benchmark models in accuracy.
A novel algorithm for multi-subband signal fusion achieves performance superior to traditional all-pole model, matrix pencil algorithm and deep-neural-network (Deep neural network (DNN)). The method uses a deep-learni...
详细信息
A novel algorithm for multi-subband signal fusion achieves performance superior to traditional all-pole model, matrix pencil algorithm and deep-neural-network (Deep neural network (DNN)). The method uses a deep-learning autoencoder more fully described as a multi-subband fusion autoencoder (MSFAE). This autoencoder comprises two parts: a multi-subband encoder and a full-band decoder. Full-band echo distance envelopes are used as training data for the full-band autoencoder, to obtain the full-band coding and the full-band decoder. Then, the multi-subband echo distance envelopes are used as training data, and the full-band coding is used as labels, to train the multi-subband encoder. Finally, the multi-subband encoder and the full-band decoder are combined to obtain the MSFAE. The multi-subband distance envelopes are input to the MSFAE to obtain the full-band distance envelopes, improving the radar distance resolution and obtaining high-resolution range profiles. In contrast with the traditional all-pole model and matrix pencil algorithm, the authors' MSFAE directly processes the information in the frequency domain, avoiding the error of pole estimation in the echo domain. In contrast with DNN, the authors' MSFAE needs only multi-subband distance envelopes as input, avoiding noise subband redundancy. The experimental results show that the fusion accuracy of MSFAE is higher than the traditional all-pole model, matrix pencil algorithm and DNN. The MSFAE has superior performance using the fusion method, even at low signal-to-noise ratio.
Seismic inversion problems often involve strong nonlinear relationships between model and data so that their misfit functions usually have many local minima. Global optimization methods are well known to be able to fi...
详细信息
Seismic inversion problems often involve strong nonlinear relationships between model and data so that their misfit functions usually have many local minima. Global optimization methods are well known to be able to find the global minimum without requiring an accurate initial model. However, when the dimensionality of model space becomes large, global optimization methods will converge slow, which seriously hinders their applications in large-dimensional seismic inversion problems. In this article, we propose a new method for large-dimensional seismic inversion based on global optimization and a machine learning technique called autoencoder. Benefiting from the dimensionality reduction characteristics of autoencoder, the proposed method converts the original large-dimensional seismic inversion problem into a low-dimensional one that can be effectively and efficiently solved by global optimization. We apply the proposed method to seismic impedance inversion problems to test its performance. We use a trace-by-trace inversion strategy, and regularization is used to guarantee the lateral continuity of the inverted model. Well-log data with accurate velocity and density are the prerequisite of the inversion strategy to work effectively. Numerical results of both synthetic and field data examples clearly demonstrate that the proposed method can converge faster and yield better inversion results compared with common methods.
This paper provides a new avenue for exploiting deep neural networks to improve physics-based simulation. Specifically, we integrate the classic Lagrangian mechanics with a deep autoencoder to accelerate elastic simul...
详细信息
This paper provides a new avenue for exploiting deep neural networks to improve physics-based simulation. Specifically, we integrate the classic Lagrangian mechanics with a deep autoencoder to accelerate elastic simulation of deformable solids. Due to the inertia effect, the dynamic equilibrium cannot be established without evaluating the second-order derivatives of the deep autoencoder network. This is beyond the capability of off-the-shelf automatic differentiation packages and algorithms, which mainly focus on the gradient evaluation. Solving the nonlinear force equilibrium is even more challenging if the standard Newton's method is to be used. This is because we need to compute a third-order derivative of the network to obtain the variational Hessian. We attack those difficulties by exploiting complex-step finite difference, coupled with reverse automatic differentiation. This strategy allows us to enjoy the convenience and accuracy of complex-step finite difference and in the meantime, to deploy complex-value perturbations as collectively as possible to save excessive network passes. With a GPU-based implementation, we are able to wield deep autoencoders (e.g., 10+ layers) with a relatively high-dimension latent space in real-time. Along this pipeline, we also design a sampling network and a weighting network to enable weight-varying Cubature integration in order to incorporate nonlinearity in the model reduction. We believe this work will inspire and benefit future research efforts in nonlinearly reduced physical simulation problems.
作者:
caliskan, AbidinBatman Univ
Fac Engn & Architecture Dept Comp Engn Batman Turkiye Batman Univ
Fac Engn & Architecture Dept Comp Engn TR-72060 Batman Turkiye
Malaria is a febrile illness caused by a parasite called plasmodium. This life-threatening disease is preventable and treatable if diagnosed early. The World Health Organization aims to reduce the global malaria incid...
详细信息
Malaria is a febrile illness caused by a parasite called plasmodium. This life-threatening disease is preventable and treatable if diagnosed early. The World Health Organization aims to reduce the global malaria incidence and death rates by at least 90% until 2030. This disease is diagnosed by visually analyzing red blood cells with a microscope by experienced radiologists. Therefore, this situation may be erroneous due to subjective interpretations. In this study, red blood cells were trained with deep learning-based convolutional neural networks to diagnose malaria, and thus, their deep features were obtained. These obtained features are also trained with autoencoder networks. Thus, the chi-square feature selection algorithm was used to obtain distinctive features. Finally, the unique feature set obtained is given as an introduction to machine learning algorithms, and then a unique diagnostic model is proposed. As a result, 100% accuracy rate was obtained. The results are promising for the diagnosis of malaria disease.
Electrical capacitance tomography (ECT) image reconstruction has developed decades and made great achievements, but there is still a need to find new theory framework to make image reconstruction results better and fa...
详细信息
Electrical capacitance tomography (ECT) image reconstruction has developed decades and made great achievements, but there is still a need to find new theory framework to make image reconstruction results better and faster. Recent years, deep learning, which is based on different series of artificial neural networks good at mapping complicated nonlinear functions, is flourishing and adopted in many fields. In this paper, a supervised autoencoder neural network is proposed to solve the image reconstruction problem of ECT, which has an encoder network and a decoder network. A simulation-based data set consisting of 40 000 pairs of instances, of which each pair of sample has a capacitance vector and corresponding permittivity distribution vector, is used to train and test the performance of the autoencoder by 10-fold cross validation. Furthermore, data with artificial noise, data regarding flow pattern not in training data set, and experimental data from a practical ECT system, are used to test the generalization ability and practicability of the network, respectively. The preliminary results show that the proposed autoencoder-based image reconstruction algorithm for ECT is of providing better reconstruction results.
Novelty detection detects outliers located at any location, such as abnormalities (i.e., far distance outliers) and novel/unobserved patterns (i.e., close distance outliers). While many novelty detection approaches ha...
详细信息
Novelty detection detects outliers located at any location, such as abnormalities (i.e., far distance outliers) and novel/unobserved patterns (i.e., close distance outliers). While many novelty detection approaches have been proposed in the literature, they generally focus on detecting one specific type of outlier, e.g., Multi-Class Open Set Recognition (MCOSR) and One-Class Novelty Detection (OCND) approaches are ap-plied for far and close distance outlier detection, respectively. However, in practice, it is difficult to mea-sure in advance whether the distance between outliers and inliers is far or close. Recent work on outlier detection at any location with a unified model has yielded mixed performance. In this paper, we pro-pose a new unified model, named Calibrated Reconstruction Based Adversarial autoencoder (CRAAE), for location agnostic outlier detection. The key idea is to integrate implicit and explicit confidence calibra-tion strategies into a reconstruction based model for building a more accurate decision boundary. We leverage the category information disentangled from feature space to calibrate the decision metric (i.e., reconstruction error) constructed in the original data space. CRAAE also adds Uniform or Dirichlet noise into the artificial outlier generation process to represent various outliers. Experimental results show that CRAAE can outperform state-of-the-art unified models (e.g., GPND) and achieve similar performance with OCND and MCOSR methods in close and far distance outlier detection, respectively.(c) 2023 Elsevier B.V. All rights reserved.
A hyperspectral image (HSI) contains hundreds of spectral bands, which provide detailed spectral information, thus offering an inherent advantage in classification. The successful launch of the Gaofen-5 and ZY-1 02D h...
详细信息
A hyperspectral image (HSI) contains hundreds of spectral bands, which provide detailed spectral information, thus offering an inherent advantage in classification. The successful launch of the Gaofen-5 and ZY-1 02D hyperspectral satellites has promoted the need for large-scale geological applications, such as mineral and lithological mapping (LM). In recent years, following the success of computer vision, deep learning methods have shown their advantage in solving the problem of hyperspectral classification. However, the combination of deep learning and HSI to solve the problem of geological mapping is insufficient. We propose a new 3D convolutional autoencoder for LM. A pixel-based and cube-based 3D convolutional neural network architecture is designed to extract spatial-spectral features. Traditional and machine learning methods are employed as competing methods, trained on two real hyperspectral datasets, and evaluated according to the overall accuracy, F1 score, and other metrics. Results indicate that the proposed method can provide convincing results for LM applications on the basis of the hyperspectral data provided by the ZY-1 02D satellite. Compared with traditional methods, the combination of deep learning and hyperspectral can provide more efficient and highly accurate results. The proposed method has better robustness than supervised learning methods and shows great promise under small sample conditions. As far as we know, this work is the first attempt to apply unsupervised spatial-spectral feature learning technology in LM applications, which is of great significance for large-scale applications. (C) The Authors. Published by SPIE under a Creative Commons Attribution 4.0 International License.
With the development of social networks, the spread of fake news brings great negative effects to people's daily life, and even causes social panic. Fake news can be regarded as an anomaly on social networks, and ...
详细信息
With the development of social networks, the spread of fake news brings great negative effects to people's daily life, and even causes social panic. Fake news can be regarded as an anomaly on social networks, and autoencoder can be used as the basic unsupervised learning method. So, an unsupervised fake news detection method based on autoencoder (UFNDA) is proposed. This paper firstly considers some forms of news in social networks, integrates the text content, images, propagation, and user information of publishing news to improve the performance of fake news detection. Next, to obtain the hidden information and internal relationship between features, Bidirectional GRU(Bi-GRU) layer and Self-Attention layer are added into the autoencoder, and then reconstruct residual to detect fake news. The experimental results compared with the existence of other four methods, on two real-world datasets, show that UFNDA obtains the more positive results.
暂无评论