检索结果-内蒙古大学图书馆

Data-driven inverse modeling with a pre-trained neural network at heterogeneous channel reservoirs

JOURNAL OF PETROLEUM SCIENCE AND ENGINEERING 2018年 170卷 785-796页

作者： Ahn, Seongin Park, Changhyup Kim, Jaejun Kang, Joe M. Samsung Heavy Ind Dept Energy Plant Researching Seongnam 13486 Gyeonggi South Korea Kangwon Natl Univ Dept Energy & Resources Engn Chunchon 24341 Kangwon South Korea Seoul Natl Univ Dept Energy Syst Engn Seoul 08826 South Korea

This paper develops a reliable and efficient data-integration method, based on artificial neural networks (ANN) incorporated with a stacked autoencoder (SAE) in a deep neural network's framework. To handle scale-different static and dynamic data of heterogeneous channel reservoirs, the workflow suggests an unsupervised pretraining process coupled with ANN-based inverse modeling. The performances of the proposed neural network, i.e. the training efficiency, the predictability of future production rates and the computing time, are compared to those with an optimal ANN and the impact of hidden neurons are discussed. The pre-trained neural network demonstrates a reliable estimation of reservoir properties with the spatial characteristics of a true channel reservoir while the ANN fails with respect to the spatial heterogeneity. The pre-trained neural network decreases the mean absolute error of future oil production rates up to 9.1% which is less than 25% of the comparison case's level, i.e. the optimal ANN model. Its efficiency is validated by a computing time 14 times faster than that of the optimal ANN workflow. The pre-trained neural network evaluates the spatial characteristics of reservoir properties and facies models in a reasonable manner and is thus able to predict the water production rates and also the breakthrough time accurately. This pre-trained neural network manifests its applicability with robustness as an efficient method to integrate static and dynamic data.

关键词： stacked autoencoder Deep neural network Pretraining Inverse model Data integration

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

stacked autoencoders Using Low-Power Accelerated Architectures for Object Recognition in Autonomous Systems

引用

NEURAL PROCESSING LETTERS 2016年第2期43卷 445-458页

作者： Maria, Joao Amaro, Joao Falcao, Gabriel Alexandre, Luis A. Univ Coimbra Inst Telecomunicacoes Dept Elect & Comp Engn P-3030290 Coimbra Portugal Univ Beira Interior Dept Informat P-6201001 Covilha Portugal Univ Beira Interior Inst Telecomunicacoes P-6201001 Covilha Portugal

This paper investigates low-energy consumption and low-power hardware models and processor architectures for performing the real-time recognition of objects in power-constrained autonomous systems and robots. Most recent developments show that convolutional deep neural networks are currently the state-of-the-art in terms of classification accuracy. In this article we propose to use of a different type of deep neural network-stacked autoencoders-and show that within a limited number of layers and nodes, for accommodating the use of low-power accelerators such as mobile GPUs and FPGAs, we are still able to achieve both classification levels not far from the state-of-the-art and a high number of processed frames per second. We present experiments using the color CIFAR-10 dataset. This enables the adaptation of the architecture to a live feed camera. Another novelty equally proposed for the first time in this work suggests that the training phase can also be performed in these low-power devices, instead of the usual approach that uses a desktop CPU or a GPU to perform this task and only runs the trained network later on the FPGA. This allows incorporating new functionalities as, for example, a robot performing runtime learning.

关键词： Deep learning Neural networks stacked autoencoder Parallel computing FPGAs Mobile GPUs OpenCL Low-power Autonomous systems

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

A Deep Learning Technique for Process Fault Detection and Diagnosis in the Presence of Incomplete Data

A Deep Learning Technique for Process Fault Detection and Di...

引用

第30届中国过程控制会议（CPCC 2019）

作者： Cen Guo Wenkai Hu Fan Yang Dexian Huang Cornell University University of Alberta Department of Automation Tsinghua University

In modern industrial processes, timely detection and diagnosis of process abnormalities are critical for monitoring process *** fault detection and diagnosis(FDD) methods have been proposed and implemented, the performance of which,however, could be drastically influenced by the common presence of incomplete or missing data in real industrial *** paper presents a new FDD approach based on an incomplete data imputation technique for process fault *** employs the modified stacked autoencoder, a deep learning structure, in the phase of incomplete data treatment, and classifies data representations rather than the imputed complete data in the phase of fault identification.A benchmark process, the Tennessee Eastman process, is employed to illustrate the effectiveness and applicability of the proposed method.

关键词： Alarm configuration deep learning fault detection and diagnosis incomplete data stacked autoencoder

来源：评论

学校读者我要写书评

暂无评论

An Empirical Study on Unsupervised Pre-training Approaches in Regression Problems 8

An Empirical Study on Unsupervised Pre-training Approaches i...

引用

8th IEEE Symposium Series on Computational Intelligence (IEEE SSCI)

作者： Saikia, Pallabi Baruah, Rashmi Dutta Indian Inst Technol Guwahati Comp Sci & Engn Dept Gauhati India

ISBN: (纸本)9781538692769

Unsupervised pre-training allows for efficient training of deep architectures. It provides a good set of initialised weights to the deep architecture that can provide better generalisation of the data. In this paper, we aim to empirically analyse the effect of different unsupervised pre-training approaches for the task of regression on different datasets. We have considered two most common pre-training methods namely deep belief network and stacked autoencoder, and compared the results with the standard training algorithm without pretraining. The models with pretraining performed better than the model without pretraining in terms of error, convergence and the prediction of pattern. The results of the experiments also show the importance of hyperparameters tuning, specially learning rate, in providing a better prediction result. This study once again confirmed the effectiveness and potential of pretraining approach in nonlinear regression problem.

关键词： Deep Architecture Multilayer perceptron Deep learning Unsupervised pre-training Deep belief network Restricted Boltzmann machine Contrastive divergence stacked autoencoder

来源：评论

学校读者我要写书评

暂无评论

FORECAST OF LOGISTICS DEMAND BASED ON GREY DEEP NEURAL NETWORK MODEL 17

FORECAST OF LOGISTICS DEMAND BASED ON GREY DEEP NEURAL NETWO...

引用

International Conference on Machine Learning and Cybernetics (ICMLC)

作者： Yuan, Wen-Jing Chen, Jian-Hua Cao, Jing-Jing Jin, Ze-Yi Wuhan Univ Technol Sch Logist Engn Wuhan 430063 Peoples R China

ISBN: (纸本)9781538652145

Aiming at the current problem of limited feature extraction ability and prediction accuracy of shallow neural networks, this paper proposes a grey deep neural network model by combining grey model and stacked autoencoder (SAE) to take advantage of their complementary merits for the particularity of logistics demand forecasting issue. Multiple grey model processes the original data first, then the output data is input into stacked autoencoder prediction (SAEP) model, applying SAE for feature representation at the bottom and extreme learning machine (ELM) for accurate prediction at the top, then the model gets a final prediction value. The proposed method is applied in the empirical research on the logistics demand in a Brazilian logistics company, and the predictive value obtains a relatively better accuracy than ordinary grey neural network model, thus the effectiveness of the proposed model has been verified.

关键词： Logistics demand prediction stacked autoencoder Grey model

来源：评论

学校读者我要写书评

暂无评论

Efficient Framework for Predicting ncRNA-Protein Interactions Based on Sequence Information by Deep Learning 14th

Efficient Framework for Predicting ncRNA-Protein Interaction...

引用

14th International Conference on Intelligent Computing (ICIC)

作者： Zhan, Zhao-Hui You, Zhu-Hong Zhou, Yong Li, Li-Ping Li, Zheng-Wei China Univ Min & Technol Sch Comp Sci & Technol Xuzhou 21116 Jiangsu Peoples R China Chinese Acad Sci Xinjiang Tech Inst Phys & Chem Urumqi 830011 Peoples R China

ISBN: (纸本)9783319959320;9783319959337

The interactions between proteins and RNA (RPIs) play a crucial role in most cellular processes such as RNA stability and translation. Although there have been many high-throughput experiments recently to detect RPIs, these experiments are largely time-consuming and labor-intensive. Therefore, it is imminent to propose an efficient computational method to predict RPIs. In this study, we put forward a novel approach for predicting protein and ncRNA interactions based on sequences information only. By employing the bi-gram probability feature extraction method and k-mer algorithm, the represent features from protein and ncRNA were extracted. To evaluate the performance of the proposed model, two widely used datasets named RPI1807 and RPI2241 were trained with the adoption of random forest classifier by using five-fold cross-validation. The experimental results with the AUC of 0.992 and 0.947 on dataset RPI1807 and RPI2241 respectively indicated the effectiveness of our experimental approach for predicting RPIs, which provided the guidance for reference for future research in the biological field.

关键词： Protein-ncRNA interaction Bi-gram Deep learning stacked autoencoder PSSM

来源：评论

学校读者我要写书评

暂无评论

Deep Learning for Integrated Analysis of Breast Cancer Subtype Specific Multi-omics Data

Deep Learning for Integrated Analysis of Breast Cancer Subty...

引用

IEEE-Region-10 Conference (IEEE TENCON)

作者： Rakshit, Somnath Saha, Indrajit Chakraborty, Subha Shankar Plewczyski, Dariusz Jalpaiguri Govt Engn Coll Dept Comp Sci & Engn Jalpaiguri India Natl Inst Tech Teachers Training & Res Dept Comp Sci & Engn Kolkata India Univ Warsaw Lab Funct & Struct Genom Ctr New Technol Warsaw Poland

ISBN: (纸本)9781538654576

Breast cancer is a deadly disease which commonly occurs all over the world and has been found to be the largest cause of cancer in females. Its detection is still a major challenge, both from a computational and biological point of views. Next Generation Sequencing (NGS) techniques have accelerated the mapping of human genomes rapidly. Involvement of advanced NGS techniques reveals that multiple genetic molecules are responsible for the cause of breast cancer and its subtypes. However, the high volume of data that is produced by the NGS techniques is difficult to study because of their high dimensionality and complexity. Thus, the integrated study of multi-omics data is one of the major challenges in medical science. This fact motivated us to study the NGS based high throughput expression data of miRNAs and mRNAs as well as Beta values of DNA Methylation of the corresponding mRNAs. In this regard, first, these datasets, together consisting of 33564 features of 305 patients in five classes viz. Luminal A, Luminal B, HER2-enriched, Basal-like and Control, are analysed in an integrated fashion using deep learning technique to classify the breast cancer subtypes properly. Second, the results of the deep learning technique are further analysed in order to identify the deeply connected features, i.e. either miRNA or mRNA or DNA Methylation, which are pivotal in the classification of breast cancer subtypes as well as play a crucial role in its formation. For this purpose, a deep learning technique, called stacked autoencoder is used to encode/transform the features into a low dimensional space, which is then fed to the five well known classifiers for classification. Moreover, the same encoded data is used to select the potential features after performing multiplication with the original data and Bonferroni correction on the p-values produced by the one-sample t-test. The results have been validated quantitatively and through biological significance analysis where oncogene

关键词： Breast cancer subtypes Deep learning Integrated analysis Multi-omics data stacked autoencoder

来源：评论

学校读者我要写书评

暂无评论

Cross-lingual sentiment classification with stacked autoencoders

引用

KNOWLEDGE AND INFORMATION SYSTEMS 2016年第1期47卷 27-44页

作者： Zhou, Guangyou Zhu, Zhiyuan He, Tingting Hu, Xiaohua Tony Cent China Normal Univ Sch Comp Wuhan 430079 Peoples R China Cent China Normal Univ Sch Comp Nat Language Proc Lab Wuhan 430079 Peoples R China Chinese Inst Elect Beijing 100036 Peoples R China Drexel Univ Coll Comp & Informat Philadelphia PA 19104 USA

Cross-lingual sentiment classification is a popular research topic in natural language processing. The fundamental challenge of cross-lingual learning stems from a lack of overlap between the feature spaces of the source language data and the target language data. In this article, we propose a new model which uses stacked autoencoders to learn language-independent high-level feature representations for the both languages in an unsupervised fashion. The proposed framework aims to force the aligned input bilingual sentences into a common latent space, and the objective function is defined by minimizing the input and output vector representations as well as the distance of the common representations in the latent space. Sentiment classifiers trained on the source language can be adapted to predict sentiment polarity of the target language with the language-independent high-level feature representations. We conduct extensive experiments on English-Chinese sentiment classification tasks of multiple data sets. Our experimental results demonstrate the efficacy of the proposed cross-lingual approach.

关键词： Sentiment classification Cross-lingual stacked autoencoder

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：