检索结果-内蒙古大学图书馆

A deep autoencoder-Based Hybrid Recommender System

INTERNATIONAL JOURNAL OF MOBILE COMPUTING AND MULTIMEDIA COMMUNICATIONS 2022年第1期13卷

作者： Bougteb, Yahya Ouhbi, Brahim Frikh, Bouchra Zemmouri, Elmoukhtar Moulay Ismail Univ ENSAM LM2I Lab Meknes Morocco Ecole Natl Sci Appl Fes Fes Morocco Moulay Ismail Univ ENSAM Ensam Morocco

Recommender systems build their suggestions on rating data, given explicitly or implicitly by users on items. These ratings create a huge sparse user-item rating matrix, which opens two challenges for researchers on the field. The first challenge is the sparsity of the rating matrix, and the second one is the scalability of the data. This article proposes a hybrid recommender system based on deep autoencoder to learn the user interests and reconstruct the missing ratings. Then, SVD++ decomposition is employed, in parallel, to hold information of correlation between different features factors. Additionally, the authors gave a deep analysis of the top-N recommender list from the user's perspective to ensure that the proposed model can be used for practical application. Experiments showed that the proposed model performed better with high-dimensional datasets and outperformed various hybrid algorithms in terms of RMSE, MAE, and in terms of the final top-N recommendation list.

关键词： deep autoencoder deep Learning Factorization Machines Learning User's Profile Recommender Systems Sparsity Problem SVD++ Top-N Recommendation List

来源：评论

学校读者我要写书评

暂无评论

Health indicator for machine condition monitoring built in the latent space of a deep autoencoder

引用

RELIABILITY ENGINEERING & SYSTEM SAFETY 2022年 224卷

作者： Gonzalez-Muniz, Ana Diaz, Ignacio Cuadrado, Abel A. Garcia-Perez, Diego Univ Oviedo Elect Engn Dept Edif Dept Oeste 2Campus Viesques S-N Gijon 33204 Spain

The construction of effective health indicators plays a key role in the engineering systems field: they reflect the degradation degree of the system under study, thus providing vital information for critical tasks ranging from anomaly detection to remaining useful life estimation, with benefits such as reduced maintenance costs, improved productivity or increased machine availability. The reconstruction error of deep autoencoders has been widely used in the literature for this purpose, but this approach does not fully exploit the hierarchical nature of deep models. Instead, we propose to take advantage of the disentangled representations of data that are available in the latent space of autoencoders, by using the latent reconstruction error as machine health indicator. We have tested our proposal on three different datasets, considering two types of autoencoders (deep autoencoder and variational autoencoder), and comparing its performance with that of state-of-the-art approaches in terms of well-known quality metrics. The results of the research demonstrate the capability of our health indicator to outperform conventional approaches, in the three datasets, and regardless of the type of autoencoder used to generate the residuals. In addition, we provide some intuition on the suitability of latent spaces for the monitoring of machinery condition.

关键词： Health indicator deep autoencoder Latent space Anomaly detection Engineering systems

来源：评论

学校读者我要写书评

暂无评论

Reconstruction of Compton Edges in Plastic Gamma Spectra Using deep autoencoder

引用

SENSORS 2020年第10期20卷 2895页

作者： Jeon, Byoungil Lee, Youhan Moon, Myungkook Kim, Jongyul Cho, Gyuseong Korea Atom Energy Res Inst Intelligent Comp Lab Daejeon 34507 South Korea Korea Adv Inst Sci & Technol Dept Nucl & Quantum Engn Daejeon 34141 South Korea Korea Atom Energy Res Inst Quantum Beam Sci Div Daejeon 34507 South Korea

Plastic scintillation detectors are widely utilized in radiation measurement because of their unique characteristics. However, they are generally used for counting applications because of the energy broadening effect and the absence of a photo peak in their spectra. To overcome their weaknesses, many studies on pseudo spectroscopy have been reported, but most of them have not been able to directly identify the energy of incident gamma rays. In this paper, we propose a method to reconstruct Compton edges in plastic gamma spectra using an artificial neural network for direct pseudo gamma spectroscopy. Spectra simulated using MCNP 6.2 software were used to generate training and validation sets. Our model was trained to reconstruct Compton edges in plastic gamma spectra. In addition, we aimed for our model to be capable of reconstructing Compton edges even for spectra having poor counting statistics by designing a dataset generation procedure. Minimum reconstructible counts for single isotopes were evaluated with metric of mean averaged percentage error as 650 for Co-60, 2000 for Cs-137, 3050 for Na-22, and 3750 for Ba-133. The performance of our model was verified using the simulated spectra measured by a PVT detector. Although our model was trained using simulation data only, it successfully reconstructed Compton edges even in measured gamma spectra with poor counting statistics.

关键词： plastic gamma spectra energy broadening correction Compton edge reconstruction deep learning deep autoencoder

来源：评论

学校读者我要写书评

暂无评论

deep-ReID: deep features and autoencoder assisted image patching strategy for person re-identification in smart cities surveillance

引用

MULTIMEDIA TOOLS AND APPLICATIONS 2024年第5期83卷 15079-15100页

作者： Khan, Samee Ullah Hussain, Tanveer Ullah, Amin Baik, Sung Wook Sejong Univ Seoul 143747 South Korea

Person Re-identification (P-ReID) task searches for true matches of a given query from a large repository of non-overlapping camera's images/videos. In smart cities surveillance, P-ReID is challenging due to variation in human's appearance, illumination affects, and difference in viewpoints. The mainstream approaches achieve P-ReID by implementing supervised learning strategies, requiring exhaustive manual annotation, which is probably erroneous due to human involvement. In addition, the employed methods use high-dimensional feature maps to identify a person, which is not a realistic approach in terms of storage resources and computational complexity. To tackle these issues, we incorporate learned features and deep autoencoder under the umbrella of a unified framework for P-ReID. First, we apply a unique image patching strategy by dividing the input image into two parts (upper and lower) and acquire learned features from fully connected layer of a pretrained Convolutional Neural Network (CNN) model for both patches. To achieve efficient and high performance, the proposed framework utilizes a self-tuned autoencoder to acquire low-dimensional representative features. The obtained features are matched with the patterns of database via cosine similarity measurement to re-identify a person's appearance. The proposed framework provides a trade-off between time complexity and accuracy, where a lightweight model can be incorporated with reduced number of autoencoder layers to obtain fast and comparatively flexible results. The major novelty of the proposed framework includes implementation of a hybrid network mechanism for P-ReID, which shows convincing real-time results and best fits for smart cities surveillance. The proposed framework is tested over several P-ReID datasets to prove its influence over the existing works with reduced computational complexity.

关键词： Smart surveillance Person re-identification deep autoencoder deep learning Learned features extraction Hybrid CNN

来源：评论

学校读者我要写书评

暂无评论

1 Unsupervised hyperspectral band selection with deep autoencoder unmixing

引用

INTERNATIONAL JOURNAL OF IMAGE AND DATA FUSION 2022年第3期13卷 244-261页

作者： Elkholy, Menna M. Mostafa, Marwa S. Ebeid, Hala M. Tolba, Mohamed Ain Shams Univ Fac Comp & Informat Sci Cairo Egypt Natl Author Remote Sensing & Space Sci Data Recept Anal & Receiving Stn Affairs Cairo Egypt

Hyperspectral imaging (HSI) is a beneficial source of information for numerous civil and military applications, but high dimensionality and strong correlation limits HSI classification performance. Band selection aims at selecting the most informative bands to minimise the computational cost and eliminate redundant information. In this paper, we propose a new unsupervised band selection approach that benefits from the current dominant stream of deep learning frameworks. The proposed approach consists of two consecutive phases: unmixing and cluster. In the unmixing phase, we utilised a nonlinear deep autoencoder to extract accurate material spectra. In the cluster phase, we calculate the variance for each obtained endmember to construct a variances vector. Then, classical K-mean was adopted to cluster the variances vectors. Finally, the optimal band subset was obtained by choosing only one spectral band for each cluster. We carried out several experiments on three hyperspectral datasets to test the feasibility and generality of the proposed approach. Experimental results indicate that the proposed approach surpasses several state-of-the-art counterparts by an average of 4% in terms of overall accuracy.

关键词： Remote sensing (RS) hyperspectral imagery (HSI) band selection (BS) nonlinear unmixing deep autoencoder dimensionality reduction (DR)

来源：评论

学校读者我要写书评

暂无评论

Vibration Recognition Based on Feature Extraction by deep autoencoder 23

Vibration Recognition Based on Feature Extraction by Deep Au...

引用

Proceedings of the 2023 International Conference on Frontiers of Artificial Intelligence and Machine Learning

作者： Qingsong Xiong Cheng Yuan Keyan Ji Chang He Hai Bei Xiong State Key Laboratory of Disaster Reduction in Civil Engineering Tongji University Shanghai China China and Department of Civil and Environmental Engineering The Hong Kong Polytechnic University Hong Kong State Key Laboratory of Disaster Reduction in Civil Engineering Tongji University Shanghai China China

ISBN: (纸本)9798400707544

Vibrational response identification of high-rise building structures under excitation of varying ambient conditions is of great significance for structural vibration control and health monitoring. Traditional strategies allocating time-frequency domain transformation or graph neural network for feature extraction cast great loss on computational efficacy and engineering applicability. Herein, this study proposes a novel vibration recognition method based on feature extraction by deep autoencoder. In this method, a deep autoencoder (AUE) is established for vibration signal reconstruction in frequency domain, on the basis of which the vibration sensitive parameters are defined and quantified based on the discrepancy between original signals and the reconstructed ones. The extracted indicators are then performed as input feature engineering of downstream linear classifiers in a supervised pattern, thereby achieving rapid identification of vibration response. To validate the capacity of the proposed method, a task of vibration response identification of a super high-rise building under typical ambient excitations was conducted. The performance results achieved a satisfactory accuracy exceeding 0.95, suggesting a powerful capability and potential applicability.

关键词： Vibration recognition deep autoencoder feature extraction signal reconstruction

来源：评论

学校读者我要写书评

暂无评论

Research on Financial Data Prediction Based on deep autoencoder

Research on Financial Data Prediction Based on Deep Autoenco...

引用

2019 2nd International Conference on Financial Management,Education and Social Science(FMESS 2019)

作者： Ziming Gao Jinchang Zhang Boyang Nan Xi'an Jiaotong-liverpool University College of Information Control China University of Mining and Technology College of Economics and Management Three Gorges University

As a revolutionary achievement in machine learning in the era of big data, deep learning has developed rapidly since its introduction, and it has created an upsurge of research and application in the Internet *** the perspective of market microstructure, the formation and change of stock prices are determined by the trading behavior of buyers and ***, the mining of high-frequency market data may have a predictive ability for future stock price *** this paper, the deep learning prediction model is trained by a large amount of historical data in the sample, and the price index of the stock index futures is predicted at 1 *** accuracy of the model outside the sample is over 73%.At the same time, based on the prediction of stock price changes by deep learning stock price forecasting model, this paper proposes the intraday trading strategy of stock index *** trading strategy has accumulated 99.6% yield since 2013, with an annualized rate of return of 77.6% and a maximum retracement of-5.86%.Through the empirical study of the high-frequency price forecasting model of stock index futures, this paper verifies the effectiveness of the machine learning tool in the big data era of deep learning in stock price forecasting, and proposes a stock index futures trading strategy based on the forecasting model, which has achieved good results.

关键词： deep Learning Stock Index Futures Trend Forecasting Trading Strategy deep autoencoder

来源：评论

学校读者我要写书评

暂无评论

Human Gait Recognition with Micro-Doppler Radar and deep autoencoder

Human Gait Recognition with Micro-Doppler Radar and Deep Aut...

引用

International Conference on Pattern Recognition

作者： Hoang Thanh Le Son Lam Phung Abdesselam Bouzerdoum School of Electrical Computer and Telecommunications Engineering University of Wollongong NSW Australia

The micro-Doppler signals from moving objects contain useful information about their motions. This paper introduces a novel approach for human gait recognition based on backscattered signals from a micro-Doppler radar. Three different signal techniques are utilized for the extraction of micro-Doppler features via time-frequency and time-scale representations. To classify the human motions into various types, this paper presents a deep autoencoder with the use of local patches extracted along the spectrogram and scalogram. The network configuration and the learning parameters of the deep autoencoder, which are considered as hyperparameters, are optimized by a Bayesian optimization algorithm. Experimental results produced by the proposed technique on real radar data show a significant improvement compared to several existing approaches.

关键词： Micro-Doppler radar Short-time fourier transform S-method Wavelet transform deep autoencoder Bayesian optimization

来源：评论

学校读者我要写书评

暂无评论

Unsupervised Facial Image Occlusion Detection with deep autoencoder

Unsupervised Facial Image Occlusion Detection with Deep Auto...

引用

2019第十一届数字图像处理国际会议

作者： Wang Xu-dong Wei Hong-quan Li Shao-mei Gao Chao Huang Rui-yang National Digital Switching System Engineering & Technological R&D Center

Face recognition techniques have been developed significantly in recent ***,recognizing faces with partial occlusion is still a challenging *** there are many works to solve the problem of obscuring the face,the occlusion is still a challenge in face *** overcome this issue,firstly we should detect the occlusion position in the facial *** construct a robust self-encoding machine to solve the occlusion detection problem in face images and uses synthetic occlusion data for *** evaluated our method under various synthetic occlusion face *** show that our method can effectively detect various types of occlusion masks in an unsupervised manner and has better robustness to the occlusion categories.

关键词： Face Recognition Occlusion Detection deep autoencoder

来源：评论

学校读者我要写书评

暂无评论

High-performance community detection in social networks using a deep transitive autoencoder

引用

INFORMATION SCIENCES 2019年 493卷 75-90页

作者： Xie, Ying Wang, Xinmei Jiang, Dan Xu, Rongbin Putian Univ Sch Informat Engn Putian Peoples R China Anhui Univ Sch Comp Sci & Technol Hefei Anhui Peoples R China

Community structure is an important characteristic of complex networks. It determines where important functions of a network are located. Recently, discovering community structure in complex networks has become a hot topic of research. However, the continuous increase in network size has made network structure more complex, and community detection has become extremely difficult in real applications. In particular, the detection results are usually not accurate enough when classical clustering methods are applied to high-dimensional data matrices. In this paper, inspired by the relationship between vertices, we design a novel and effective network adjacency matrix transformation method to describe vertices' similarity in the network topology. On this basis, we propose a framework to extract nonlinear features: community detection with deep transitive autoencoder (CDDTA). This framework can obtain powerful nonlinear features of a real network to make community detection algorithms perform excellently in practice. We further incorporate unsupervised transfer learning into the CDDTA (Transfer-CDDTA) by minimizing the Kullback-Leibler divergence of embedded instances, to discover powerful low-dimensional representations. Finally, we propose a new training strategy and optimization method for our algorithm. Extensive experimental results indicate that our new framework can ensure good performance on both real-world networks and artificial benchmark networks, which outperforms most of the state-of-the-art methods for community detection in social networks. (C) 2019 Elsevier Inc. All rights reserved.

关键词： Community detection Complex networks deep autoencoder Unsupervised transfer learning

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：