检索结果-内蒙古大学图书馆

18th International Conference on Man-Machine-Environment System Engineering (MMESE)

作者： Lan, Wei Liu, Yixin Qi, Zhang Song, Shimin He, Chun Wang, Lijing Li, Ke Univ Aeronaut & Astronaut Beijing 100191 Peoples R China Beihang Univ Sch Aeronaut Sci & Engn Fundamental Sci Ergon & Environm Control Lab Beijing 100191 Peoples R China China Acad Space Technol Beijing 100094 Peoples R China

ISBN: (纸本)9789811324819;9789811324802

Based on deep learning, a multi-classification algorithm network is designed for the large amount of data generated in spacecraft test. In the algorithm, the initial offsets and weights of a multi-layer neural network are initialized using an auto-encoder method. The initialized parameters are monitored by the gradient descent method to make the dimension data more separable. Many shortcomings of traditional algorithms can be effectively overcome using this algorithm. For example, the storage space can be reduced and the calculation time can be saved when the data is large or complex. Expert knowledge of the spacecraft health management platform can be provided through the study of measured data. Experimental data shows that the depth learning algorithm which is based on SAE has higher accuracy in spacecraft multi-class signal testing.

关键词： PHM Deep learning auto-encoder Pattern recognition Data compression Deep belief network

来源：评论

学校读者我要写书评

暂无评论

A Deep Neural Network Based Hierarchical Multi-Label Classifier for Protein Function Prediction

A Deep Neural Network Based Hierarchical Multi-Label Classif...

引用

8th International Conference on Computer, Information and Telecommunication Systems (CITS)

作者： Yuan, Xin Li, Weite Lin, Kui Hu, Jinglu Waseda Univ Grad Sch Informat Prod & Syst 2-7 Hibikino Kitakyushu Fukuoka 8080135 Japan Beijing Normal Univ Coll Life Sci 19 Xinjiekou Outer St Beijing 100875 Peoples R China

ISBN: (纸本)9781538640883

Hierarchical classification approaches have been shown to be effective for protein function prediction problem. Traditionally, a set of simple classifiers are used for each hierarchical level separately. In this paper, we introduce a deep neural network (DNN) model with multiple heads and multiple ends to realize the whole set of classifiers. The DNN model consists of three parts: the body part, the multi-end part and the multi-head part. The body part is a deep multilayer perceptron (MLP) shared by different classifiers for feature mapping. The multi-end part performs feature fusion transforming the input vectors of different classifiers to feature vectors with the same length so as to share the feature mapping part. The multi-head part is a set of linear multi-label classifiers. By sharing a deep MLP with multiple classifiers, we are able to construct more powerful classifiers for each level with limited training samples, and expecting to have better classification performance. Experiment results on benchmark datasets show that the proposed method significantly outperforms these traditional methods.

关键词： Hierarchical Multi-Label Classification Protein Function Prediction Deep Neural Network Feature Fusion auto-encoder

来源：评论

学校读者我要写书评

暂无评论

Control System Response Improvement via Denoising Using Deep Neural Networks 10

Control System Response Improvement via Denoising Using Deep...

引用

IEEE 10th Annual Ubiquitous Computing, Electronics and Mobile Communication Conference (UEMCON)

作者： Fathi, Kiavash Mahdavi, Mehdi Isfahan Univ Technol Dept Elect & Comp Engn Esfahan Iran

ISBN: (纸本)9781728138855

Noise is an inseparable part of control systems. Every sensor reading used for determining the state of a control system is corrupted with noise, therefor increasing the signal to noise ratio of sensor readings can significantly improve the performance of systems. The proposed filter in this paper captures the underlying probability distribution of the noise-free input signal in the training stage and therefor is capable of refining the corrupted input signal regardless of the distribution of the added noise. In order to acquire better data-driven based results in the suggested approach, it has been decided to use different neural network structures and stack these layers to form a hybrid multilayer filter. The properties of the stacked neural network sublayers guarantee a robust and general solution for the intended purpose. The key elements of the proposed filter are two autoencoders, a dense neural network and a convolutional layer. Effects of every sublayer on the corrupted input signal, along with fine-tuning of these sublayers are discussed in detail. Afterwards in order to assess the generality and the robustness of the method, the proposed filter is exposed to a non-Gaussian noise. Finally, the proposed filter is tested on a linear and a nonlinear system. The comparison between systems' output to the reconstructed signal with that of the original noise-free signal, suggests the substantial improvement in the performance of the given systems. This improvement is in terms of systems' output resemblance to the reference noise-free output signal when the reconstructed signal is applied to the systems.

关键词： Deep learning auto-encoder Dense neural network Control system Sensor reading Signal Processing Denoising

来源：评论

学校读者我要写书评

暂无评论

A Batch-normalized Deep Neural Networks and its application in bearing fault diagnosis 11

A Batch-normalized Deep Neural Networks and its application ...

引用

11th International Conference on Intelligent Human-Machine Systems and Cybernetics (IHMSC)

作者： Zheng, Jiaqi Sun, Hongjun Wang, Xiaojing Liu, Jun Zhu, Caizhi Shanghai Univ Sch Mech Engn & Automat 99 Shangda Rd Shanghai 200444 Peoples R China

ISBN: (纸本)9781728118598

At present, in the field of fault diagnosis, deep learning has shown state-of-the-art performance in processing mechanical big data. This paper studies the deep neural networks(DNN) model based on auto-encoder, which has high performance in bearing fault diagnosis. However, the traditional structure of stacked auto-encoders has the problem of internal covariant transfer, that inhibits the training efficiency and generalization ability of the network. To overcome the aforementioned deficiency and further explore the performance of DNN, a batch normalization layer is employed in the fully connected layer of the DNN during training, so the network can obtain the stable distribution of activation values. Therefore, this paper proposes a new intelligent diagnosis method named batch normalization deep neural networks(BN-DNN). Finally, the experimental results show that: (1) The performance of BN-DNN is better than DNN. (2) BN-DNN can directly process the raw vibration signals, and the diagnostic accuracy can be maintained above 99% under different working conditions.

关键词： Deep learning fault diagnosis auto-encoder batch normalization

来源：评论

学校读者我要写书评

暂无评论

Regularized Learning of Neural Network with Application to Sparse PCA 18th

Regularized Learning of Neural Network with Application to S...

引用

18th International Conference on Artificial Intelligence and Soft Computing (ICAISC)

作者： Rodziewicz-Bielewicz, Jan Klimaszewski, Jacek Korzen, Marcin West Pomeranian Univ Technol Szczecin Fac Comp Sci & Informat Technol Zolnierska 49 St PL-71210 Szczecin Poland

ISBN: (纸本)9783030209124;9783030209117

The paper presents an implementation of the regularized two-layer neural network and its application to finding sparse components. The main part of the paper concerns the learning of the sparse regularized neural network and its use as auto-encoder. A process of learning of the neural network with non-convex optimization criterion is reduced to convex optimization with constraints in an extended domain. This approach is compared with the dictionary learning procedure. The experimental part presents the comparison of our implementation and the SparsePCA procedure from the Scikit-learn package on different data sets. As a quality of the solution during experiments we take into account: the time of learning, sparsity, and quality of reconstruction. The experiments show that our approach can be competitive when a higher sparsity is needed, and in the case of a large number of attributes relative to the number of instances.

关键词： Neural network Sparsity l(1) penalty Sparse PCA auto-encoder

来源：评论

学校读者我要写书评

暂无评论

Data Fusion-Based Network Anomaly Detection towards Evidence Theory 6

Data Fusion-Based Network Anomaly Detection towards Evidence...

引用

6th National-Foundation-for Science-and-Technology-Development (NAFOSTED) Conference on Information and Computer Science (NICS)

作者： Cong Thanh Bui Van Loi Cao Minh Hoang Quang Uy Nguyen Hitech Telecommun Ctr Commun Command Software R&D Dept Hanoi Vietnam Le Quy Don Tech Univ Fac Informat Technol Hanoi Vietnam Inst Sci Technol & Innovat Hanoi Vietnam

ISBN: (纸本)9781728151632

We propose a fusion model of deep neural networks and traditional algorithms for anomaly detection. The proposed model inherits the advantages of both these methods to create a robust anomaly detection algorithm. We employ the Dempster-Shafer theory (D-S) of Evidence, a very reliable and flexible data fusion technique, to form a fusion-based network anomaly detection (FuseNAD) by applying a basic probability assignment (BPA) function and modifying the D-S theory's rule. FuseNAD fuses four anomaly detection methods consisting of a deep learning technique, namely Shrink auto-encoder, and three traditional ones such as One-class Support Vector Machine (OCSVM), Kernel Density Estimation (KDE) and Local Outlier Factor (LOF). The experimental results show increases in detection rate and overall accuracy in comparison to the individuals on several public network anomaly detection datasets.

关键词： Deep learning auto-encoder Anomaly Detection D-S Theory Data Fusion

来源：评论

学校读者我要写书评

暂无评论

Extracting Word Embeddings via Joint Learning of Syntagmatic and Paradigmatic Structure 19

Extracting Word Embeddings via Joint Learning of Syntagmatic...

引用

Proceedings of the 2019 2nd International Conference on Algorithms, Computing and Artificial Intelligence

作者： Chao Wei Lijun Zhu Juncheng Wang Jiaoxiang Shi Zheng Wang Liang Chen National Engineering Research Center of Science and Technology Information Institute of Science and Technical Information of China Beijing China National Engineering Research Center of Science and Technology Information Institute of Science and Technical Information of China City State Country

ISBN: (纸本)9781450372619

In this work, we explore the latent distributional semantic between each given word and its context in a way of reconstruction, and propose an auto-encoder architecture to encourage the model pay attention to the interactions of Syntagmatic and Paradigmatic structure of context (SPC). In particular, SPC represent the syntagmatic words of the context through a vocabulary co-occurrence matrix, and extract a likely embedding for a better joint predicting of all paradigmatic words to fill in the input context. During the error back propagation, our model focus on the salient statistical structure by allowing a sub-network to approximate the nonzero sub-matrix. Finally, we trained our model on a public Wikipedia corpus and evaluated on word analogy, word similarity tasks and documents clustering and classification. The results show that our method yields almost 4% improvement on word analogy and nearly 3% improvement on word similarity tasks compared to state-of-the-art methods. The evidences demonstrate that SPC outperform baseline and the state-of-the-art methods on all tasks.

关键词： auto-encoder

来源：评论

学校读者我要写书评

暂无评论

Impact of Feature Representation for Imitation Learning in autonomous Drive

Impact of Feature Representation for Imitation Learning in A...

引用

作者： MALIN DAHL Chalmers University of Technology

学位级别：硕士

autonomous drive in complex traffic scenarios is a demanding task to solve. With high-dimensional input data available, problems related to redundancy and irrele- vance are often implicated, hence determining what features bring the most useful information is of vital importance. The purpose of this thesis is to investigate how different dimensionality reduction methods affect the driving performance and how to determine what features are most relevant. Specifically, these questions were studied in a simulated environment where a car is manoeuvred using deep neural networks through a sequence of signalised intersec- tions. Four different dimensionality reduction methods have been studied: choice of features based on reason, Principal Component Analysis, auto-encoders and Integrated encoders. The results showed that the models which used a feature representation based on reason were shown to perform best. Also, the weight distributions of a model using all available features indicated that influential features may be partially identified by studying the spread of the weights. Therefore, an approach is proposed where the choice of features should be based on reason as well as a study of the features' respective set of weights. In conclusion, es- tablishing the most relevant feature representation is important since it may benefit the training of the models.

关键词： Feature Representation Deep Neural Networks Imitation Learning Dimensionality Reduction Principal Component Analysis auto-encoder autonomous Drive

来源：评论

学校读者我要写书评

暂无评论

Research on water quality prediction method based on AE-LSTM

Research on water quality prediction method based on AE-LSTM

引用

automation, Control and Robotics Engineering (CACRE), International Conference on

作者： Huiqing Zhang Kemei Jin School of Automation Department of Informatics Beijing University of Technology

ISBN: (数字)9781728198880

ISBN: (纸本)9781728198897

Aiming at the traditional prediction methods of related parameters that affect water quality, they usually only consider the temporal characteristics of the related parameters of water quality and ignore the problem that water quality changes are multivariate related, A prediction method of spatiotemporal correlation water quality parameters based on automatic encoder(AE) dimensionality reduction and long and short time memory(LSTM) neural network is proposed. Firstly, considering that water quality parameter changes have obvious time characteristics, a time series prediction model of water quality parameters is established based on LSTM. Secondly, considering that the water quality changes have multiple correlations, the upstream water quality will also affect the downstream water quality. If all the water quality parameters of the upstream station are added to the prediction model, redundant features will reduce the accuracy of parameter prediction. Therefore, the automatic encoder is used to reduce the dimensionality of the relevant parameters. Finally, the data set of Lang fang Water Quality automatic Monitoring Station is applied to monitor the effectiveness of the method. By predicting the concentration of total phosphorus(TP) and total nitrogen(TN), the method is found to have better prediction accuracy and robustness.

关键词： long-short-term memory network auto-encoder water quality prediction feature extraction multivariate correlation

来源：评论

学校读者我要写书评

暂无评论

LMAE: A large margin auto-encoders for classification

引用

SIGNAL PROCESSING 2017年 141卷 137-143页

作者： Liu, Weifeng Ma, Tengzhou Xie, Qiangsheng Tao, Dapeng Cheng, Jun China Univ Petr East China Qingdao Peoples R China Shandong Inst Food & Drug Control Jinan Shandong Peoples R China Yunnan Univ Kunming Yunnan Peoples R China Chinese Acad Sci Shenzhen Inst Adv Technol Shenzhen Peoples R China

auto-encoders, as one representative deep learning method, has demonstrated to achieve superior performance in many applications. Hence, it is drawing more and more attentions and variants of auto encoders have been reported including Contractive auto-encoders, Denoising auto-encoders, Sparse auto encoders and Nonnegativity Constraints auto-encoders. Recently, a Discriminative auto-encoders is reported to improve the performance by considering the within class and between class information. In this paper, we propose the Large Margin auto-encoders (LMAE) to further boost the discriminability by enforcing different class samples to be large marginally distributed in hidden feature space. Particularly, we stack the single-layer LMAE to construct a deep neural network to learn proper features. And finally we put these features into a softmax classifier for classification. Extensive experiments are conducted on the MNIST dataset and the CIFAR-10 dataset for classification respectively. The experimental results demonstrate that the proposed LMAE outperforms the traditional auto-encoders algorithm. (C) 2017 Elsevier B.V. All rights reserved.

关键词： auto-encoder Large Margin kNN Classification

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：