检索结果-内蒙古大学图书馆

On deep ensemble CNN-SAE based novel agro-market price forecasting

EVOLUTIONARY INTELLIGENCE 2021年第2期14卷 851-862页

作者： Parida, Nirjharinee Mishra, Debahuti Das, Kaberi Rout, Narendra Kumar Panda, Ganapati Siksha O Anusandhan Deemed Univ Dept Comp Sci Engn Bhubaneswar Odisha India Natl Inst Technol Dept Comp Applicat GE Rd Chhatisgarh 492010 CG India CV Raman Global Univ Dept Elect & Telecommun Engn Bhubaneswar 752054 Odisha India

The prices of agro-commodities are highly volatile. Hence it is a challenge to the farmers to ensure fair and remunerative prices of these commodities. As a result, there is a need for prediction of agro market price appropriately. The closing price prediction of one soft commodity product, Cotton 29 mm and one agro-commodity product, Guar gum are chosen. The existing reported methods exhibit poor prediction performance. To alleviate this problem, the current investigation is undertaken for better prediction of closing prices. The deep ensemble approach using convolutional neural network (CNN) and stacked autoencoder (SAE) is employed to improve the prediction performance. For the ensemble strategy, the weights are optimized using three bio-inspired techniques such as genetic algorithm (GA), particle swarm optimization (PSO) and spider monkey optimization (SMO). Eighteen attributes relating to the closing price of each products are considered as input to the proposed models. The simulation based experimental results demonstrate the following contribution of the paper. Firstly, it is observed that CNN outperforms the SAE model in terms of short range prediction and vice versa for long range prediction. Secondly, the prediction performance of all the three ensemble models has been determined. Thirdly, out of three ensemble models, ensemble-SMO (ESMO) shows the best prediction performance in terms of mean square error and coefficient of multiple determination (R-2). It is then followed by ensemble-PSO and ensemble-GA respectively. The performance of proposed best ESMO is compared with the Grey wolf optimization based multiquadratic kernel KELM model (GWO-KELM) and it is observed that the proposed ESMO outperforms the GWO-KELM model.

关键词： Deep learning Convolutional neural network stacked autoencoder Ensemble learning Spider monkey optimization Genetic algorithm Particle swarm optimization R-2(coefficient of multiple determination)

来源：评论

学校读者我要写书评

暂无评论

Deep Learning Technique-Based Steering of Autonomous Car

引用

INTERNATIONAL JOURNAL OF COMPUTATIONAL INTELLIGENCE AND APPLICATIONS 2018年第2期17卷

作者： Yang, Yiqin Wu, Zhe Xu, Qingyang Yan, Fabao Shandong Univ Sch Mech Elect & Informat Engn 180 Wenhua Xilu Weihai 264209 Shandong Peoples R China

Deep neural network (DNN) has many advantages. Autonomous driving has become a popular topic now. In this paper, an improved stack autoencoder based on the deep learning techniques is proposed to learn the driving characteristics of an autonomous car. These techniques realize the input data adjustment and solving diffusion gradient problem. A Raspberry Pi and a camera module are mounted on the top of the car. The camera module provides the images needed for training the DNN. There are two stages in the training. In the pre-training process, an improved autoencoder is trained by the unsupervised learning mechanism, and the characterization of the track is extracted. In the fine-tuning stage, the whole network is trained according to the labeled data, and then this model learns the driving characteristics better according to the samples. In the experimental stage, the car will predict the action of the car by the trained model in the autonomous mode. The experiment exhibits the effectiveness of the proposed model. Compared with the traditional neural network, the improved stack autoencoder has a better generalization ability and faster convergence speed.

关键词： Raspberry Pi batch normalization self-driving car stacked autoencoder deep learning

来源：评论

学校读者我要写书评

暂无评论

NOTE: non-parametric oversampling technique for explainable credit scoring

引用

SCIENTIFIC REPORTS 2024年第1期14卷 1-18页

作者： Han, Seongil Jung, Haemin Yoo, Paul D. Cali, Andrea Provetti, Alessandro Univ London Birkbeck Coll Sch Comp & Math Sci London England Korea Natl Univ Transportat Dept Ind & Management Engn Chungju South Korea

Credit scoring models are critical for financial institutions to assess borrower risk and maintain profitability. Although machine learning models have improved credit scoring accuracy, imbalanced class distributions remain a major challenge. The widely used Synthetic Minority Oversampling TEchnique (SMOTE) struggles with high-dimensional, non-linear data and may introduce noise through class overlap. Generative Adversarial Networks (GANs) have emerged as an alternative, offering the ability to model complex data distributions. Conditional Wasserstein GANs (cWGANs) have shown promise in handling both numerical and categorical features in credit scoring datasets. However, research on extracting latent features from non-linear data and improving model explainability remains limited. To address these challenges, this paper introduces the Non-parametric Oversampling Technique for Explainable credit scoring (NOTE). The NOTE offers a unified approach that integrates a Non-parametric stacked autoencoder (NSA) for capturing non-linear latent features, cWGAN for oversampling the minority class, and a classification process designed to enhance explainability. The experimental results demonstrate that NOTE surpasses state-of-the-art oversampling techniques by improving classification accuracy and model stability, particularly in non-linear and imbalanced credit scoring datasets, while also enhancing the explainability of the results.

关键词： Conditional Wasserstein generative adversarial networks stacked autoencoder Explainable AI Imbalanced class Oversampling Credit scoring

来源：评论

学校读者我要写书评

暂无评论

Hyperspectral image classification using multi-task feature leverage with multi-variant deep learning

引用

EARTH SCIENCE INFORMATICS 2020年第4期13卷 1093-1102页

作者： Jayapriya, Kalyanakumar Jacob, Israel Jeena Darney, Paulraj Ebby Vin Solut Tirunelveli India GITAM Sch Technol Dept Comp Sci & Engn Bengaluru India SCAD Coll Engn & Technol Dept Elect Engn Tirunelveli India

Deep learning framework aids the researchers in learning different application areas to a greater extent. Deep learning framework is preferred over machine learning since it helps to learn the input from end to end, whereas latter one require the inputs to be cut into pieces according to the need. This paper proposes a Multi-Variant Deep Learning framework for learning and classifying the hyperspectral images. Multi-task Feature Leverage is incorporated by doing two-ordered feature extraction. The first order feature extraction was done by using Two Dimensional Empirical Wavelet Transforms (2D-EWT) and the second-order feature extraction was done by using stacked autoencoder (SAE) and Convolutional Neural Network (CNN) for the approximation image of 2D-EWT. Because of the possibility of working on prominent feature, the proposed work uses approximation image than the raw image. The classification was carried out by using Random Forest (RF), Multi- Support Vector Machine (MSVM) and Extreme learning machine (ELM).

关键词： Deep learning Random forest Multi-support vector machine Extreme learning machine stacked autoencoder Convolutional neural network

来源：评论

学校读者我要写书评

暂无评论

Classification of hyperspectral images by deep learning of spectral-spatial features

引用

ARABIAN JOURNAL OF GEOSCIENCES 2020年第12期13卷 1-14页

作者： Ding, Haiyong Xu, Luming Wu, Yue Shi, Wenzhong Nanjing Univ Informat Sci & Technol Sch Remote Sensing & Geomat Engn Nanjing 210044 Peoples R China Nanjing Univ Informat Sci & Technol Sch Geog Sci Nanjing 210044 Peoples R China Hong Kong Polytech Univ Dept Land Surveying & Geoinformat Hong Kong Peoples R China

Creating accurate land use and land cover maps using remote sensing images is one of the most important applications of remotely sensed data. Abundant spectral information in hyperspectral images (HSI) makes it possible to distinguish materials that would not be distinguishable by multi-spectral sensors. Spectral and spatial information from HSI is of primary importance for image classification. In this study, a hybrid stacked autoencoder (SAE) architecture and support vector machine (SVM) classifier was constructed to classify the HSI. The SAE architecture is constituted by stacking a multiple autoencoder (AE) deep learning network that consists in the encoder and decoder process. Spatial features in a neighbor region extracted from the principal component analysis (PCA) and the texture feature extracted from the gray-level cooccurrence matrix (GLCM) were fed into the classifier. It was found that the best result was from the combination of GLCM texture feature, PCA spatial feature, and spectral feature. Meanwhile, the representative features derived from SAE deep learning network were better than the original features. It reminded us that extracting the representative features from hyperspectral images is a key step of improving classification accuracy.

关键词： stacked autoencoder Deep learning Hyperspectral image classification Spatial features

来源：评论

学校读者我要写书评

暂无评论

WiFi-Sensing Based Person-to-Person Distance Estimation Using Deep Learning 24

WiFi-Sensing Based Person-to-Person Distance Estimation Usin...

引用

24th IEEE International Conference on Parallel and Distributed Systems (ICPADS)

作者： Liu, Wenping Jia, Yufu Jiang, Guoyin Jiang, Hongbo Wu, Fan Lv, Zhicheng Hubei Univ Econ Sch IMS Wuhan Hubei Peoples R China Hunan Univ Coll CSEE Changsha Hunan Peoples R China Univ Elect Sci & Tech China Sch PA Chengdu Sichuan Peoples R China Huazhong Univ Sci & Technol Sch EIC Wuhan Hubei Peoples R China Hubei Univ Econ Sch IE Wuhan Hubei Peoples R China

ISBN: (纸本)9781538673089

Accurately estimating the distance between persons with COTS mobile devices can benefit many applications (e.g., group activity analysis, indoor navigation, etc.). In this paper we present WiDE, a deep learning-based system for estimating person-to-person distance based on surrounding WiFi signals. Specifically, WiDE has two phases: offline learning and on-line prediction. In offline learning phase, we apply a stacked autoencoder (SAE) for pre-training the weights of a deep neural network (DNN), and establish a DNN-based classifier for predicting between-person discretized distance and corridor identity using WiFi signals. During online prediction phase, based on the trained DNN with the SAE and newly uploaded WiFi information, we estimate the corridor identities and the distance between pairwise persons. We validate our system by conducting extensive experiments in a three-floor campus building, and the results show that WiDE achieves the corridor identification accuracy over 98% and the median ranging error of 0.9m and 3.0m for two persons on the same corridor and on different corridors, respectively, which outperforms the state-of-the-art proximity inferring system [1].

关键词： Distance Estimation COTS Device Deep Neural Network stacked autoencoder

来源：评论

学校读者我要写书评

暂无评论

stacked Auto-Encoder for Scalable Indoor Localization in Wireless Sensor Networks 15

Stacked Auto-Encoder for Scalable Indoor Localization in Wir...

引用

15th IEEE International Wireless Communications and Mobile Computing Conference (IEEE IWCMC)

作者： BelMannoubi, Souad Touati, Haifa Snoussi, Hichem Univ Gabes Fac Sci Gabes Hatem Bettaher IResCoMath Res Unit Gabes Tunisia Univ Technol Troyes CNRS FRE 2848 ICD Troyes France

ISBN: (纸本)9781538677476

In this paper, we propose a Deep Neural Network model based on WiFi-fingerprinting to improve the accuracy of zone location in a multi-building, multi-floor indoor environment. The proposed model is presented as a stacked autoencoder (SAE) to allow efficient reduction of the feature space in order to achieve robust and precise classification. The multi-label classification is used to simplify and reduce the complexity of the learning classification task during the training phase. To achieve a hierarchical classification, we applied an argmax function on the multi-label output to convert the multi-label classification into multi-class classification ones to estimate the building, the floor and the zone identifier. Experimental results show that the proposed model achieves an accuracy of 100% for building, 99.66% for floor and 83.47% for zone location with a test time that does not exceed 10.21s.

关键词： Zoning Localization WiFi-Fingerprint Deep Neural Network stacked autoencoder Multi-Class Multi-Label

来源：评论

学校读者我要写书评

暂无评论

LESION DETECTION USING T1-WEIGHTED MRI: A NEW APPROACH BASED ON FUNCTIONAL CORTICAL ROIS 24

LESION DETECTION USING T1-WEIGHTED MRI: A NEW APPROACH BASED...

引用

24th IEEE International Conference on Image Processing (ICIP)

作者： Guo, Dazhou Zheng, Kang Wang, Song Univ South Carolina Dept Comp Sci & Engn 301 Main St Columbia SC 29208 USA

ISBN: (纸本)9781509021758

Accurate and precise detection of brain lesions on MR images is important for relating lesion locations to impaired behaviors. In this paper, we propose a method to detect lesion voxels on each functional cortical ROI (Region of Interest) independently using only T1-weighted MR images (T1-MRI). In contrast to existing automatic lesion detection methods, which typically detect lesion voxels on the whole MR image or on gray matter (GM)/white matter (WM), we show that the proposed functional cortical ROI based method can lead to better lesion-detection performance. We evaluate the proposed method using an in-house dataset with 60 chronic stroke patients. Using leave-one-subject-out cross validation, the proposed method can achieve an average Dice coefficient of 0.74 +/- 0.11 and outperform three state-of-the-art methods by more than 0.05.

关键词： Lesion Detection Functional Cortical ROIs T1-weighted MRI stacked autoencoder

来源：评论

学校读者我要写书评

暂无评论

Improve Traffic Prediction Using Accident Embedding on Ensemble Deep Neural Networks 11

Improve Traffic Prediction Using Accident Embedding on Ensem...

引用

11th Annual International Conference on Knowledge and Smart Technology (KST)

作者： Liyong, Wanida Vateekul, Peerapon Chulalongkorn Univ Fac Engn Dept Comp Engn Big Data Analyt & IoT Ctr CUBIC Bangkok Thailand

ISBN: (纸本)9781538675120

Traffic Prediction in a large-scale network has become an important topic of an intelligent transportation system (ITS). The problem is challenging due to various nonlinear temporal and difficulty for longer-step ahead prediction. The recurring congestion on rush-hour and non-recurring congestion on accidents are the main patterns. While most of the previously proposed techniques focus on rush-hour, unexpected accidents that affect local traffic were not considered in these works. To improve the prediction, we propose an ensemble deep neural network to archive the recurring and non-recurring congestion together combining CNN and LSTM architectures. Furthermore, the accident information is utilized and embedded using stacked autoencoder. An experiment was conducted on two data sets: England's Highways and Thailand's Expressways. The results showed that a combination of CNN and LSTM can be improved by using an embedded vector of accident information.

关键词： Traffic Prediction Convolutional Neural Networks Long Short-Term Memory stacked autoencoder Ensemble

来源：评论

学校读者我要写书评

暂无评论

A Novel Feature Representation Method Based on Deep Neural Networks for Gear Fault Diagnosis 8

A Novel Feature Representation Method Based on Deep Neural N...

引用

2017 Prognostics and System Health Management Conference (PHM-Harbin)

作者： Wang, Jinrui Li, Shunming Jiang, Xingxing Xin, Yu Nanjing Univ Aeronaut & Astronaut Coll Energy & Power Engn Nanjing Jiangsu Peoples R China Soochow Univ Sch Urban Rail Transpotat Suzhou Peoples R China

ISBN: (纸本)9781538603703

Intelligent diagnosis method has been a hot topic in the prognostics and fault diagnosis of rotating machinery. A deep neural network (DNN) method for gear fault diagnosis based on stacked autoencode (SAE) and softmax regression is presented in this work. SAE is first utilized to extract features from the frequency spectra of vibration signal. And then the learned features are used to train softmax regression classifier for identifying gear faults. The diagnosis results validate that the method is feasible and is able to acquire superior classification performance. With the increasing layers of DNN, the learned features of each hidden layer are becoming more and more robust for classification. In this paper, a novel view on the visualization of the learned features is displayed to evaluate the DNN-based fault diagnosis method.

关键词： fault diagnosis stacked autoencoder softmax regression visualization

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：