检索结果-内蒙古大学图书馆

Deep structural clustering for single-cell RNA-seq data jointly through autoencoder and graph neural network

BRIEFINGS IN BIOINFORMATICS 2022年第2期23卷 bbac018-bbac018页

作者： Gan, Yanglan Huang, Xingyu Zou, Guobing Zhou, Shuigeng Guan, Jihong Donghua Univ Sch Comp Sci & Technol Shanghai Peoples R China Washington Univ Dept Comp Sci & Engn St Louis MO 63110 USA Shanghai Univ Dept Comp Sci & Technol Shanghai Peoples R China Fudan Univ Sch Comp Sci Shanghai Peoples R China Tongji Univ Dept Comp Sci & Technol Shanghai 200092 Peoples R China

Single-cell RNA sequencing (scRNA-seq) permits researchers to study the complex mechanisms of cell heterogeneity and diversity. Unsupervised clustering is of central importance for the analysis of the scRNA-seq data, as it can be used to identify putative cell types. However, due to noise impacts, high dimensionality and pervasive dropout events, clustering analysis of scRNA-seq data remains a computational challenge. Here, we propose a new deep structural clustering method for scRNA-seq data, named scDSC, which integrate the structural information into deep clustering of single cells. The proposed scDSC consists of a Zero-Inflated Negative Binomial (ZINB) model-based autoencoder, a graph neural network (GNN) module and a mutual-supervised module. To learn the data representation from the sparse and zero-inflated scRNA-seq data, we add a ZINB model to the basic autoencoder. The GNN module is introduced to capture the structural information among cells. By joining the ZINB-based autoencoder with the GNN module, the model transfers the data representation learned by autoencoder to the corresponding GNN layer. Furthermore, we adopt a mutual supervised strategy to unify these two different deep neural architectures and to guide the clustering task. Extensive experimental results on six real scRNA-seq datasets demonstrate that scDSC outperforms state-of-the-art methods in terms of clustering accuracy and scalability. Our method scDSC is implemented in Python using the Pytorch machine-learning library, and it is freely available at .

关键词： deep clustering scRNA-Seq ZINB model autoencoder graph neural network

来源：评论

学校读者我要写书评

暂无评论

autoencoder based Communication System using Multi-Dimensional Constellations 35

Autoencoder based Communication System using Multi-Dimension...

引用

35th International Technical Conference on Circuits/Systems, Computers and Communications (ITC-CSCC)

作者： Lee, Hyungje Eom, Chahyeon Lee, Hyojin Lee, Chungyong Yonsei Univ Sch Elect & Elect Engn 50 Yonsei Ro Seoul South Korea Samsung Res Samsung Elect 56 Sungchon Gil Seoul South Korea

ISBN: (纸本)9784885523281

This paper proposes an autoencoder based multiple-input multiple-output (MIMO) communication system. The proposed autoencoder learns and optimizes for only line of sight (LOS) component of Rician channel. In addition, we adopt multi-dimensional constellation (MDC) in autoencoder, where it is obtained during learning process of autoencoder by adjusting hyperparameter. Simulation results show that our proposed autoencoder using MDC achieves better symbol error rate (SER) performance compared to conventional communication system which uses quadrature amplitude modulation (QAM) constellation. Furthermore, we confirmed that although proposed autoencoder is learned for only LOS component, it can be applied to random Rician flat fading channels with fading components and channel variation terms.

关键词： autoencoder multiple-input multiple-output multi-dimensional constellation Rician flat fading channel

来源：评论

学校读者我要写书评

暂无评论

autoencoder-based outlier detection for sparse, high dimensional data 8

Autoencoder-based outlier detection for sparse, high dimensi...

引用

8th IEEE International Conference on Big Data (Big Data)

作者： Chen, Wanghu Li, Huijun Li, Jing Arshad, Ali Northwest Normal Univ Inst Comp Sci & Engn Lanzhou 730070 Peoples R China

ISBN: (纸本)9781728162515

Outlier detection is essential in many data mining tasks. For high-dimensional data, its outlier detection often faces two challenges caused by sparse spatial distribution of data and big difficulties to get enough class labels. Therefore, it is valuable to explore a simpler and more effective approach to unsupervised outlier detection. In this paper, focusing on high-dimensional sparse data, an unsupervised outlier detection approach based on autoencoders and Robust PCA is proposed. Because Robust PAC has greater advantages in feature extraction of high-dimensional data and autoencoder has powerful capabilities in the reconstruction of normal data, the proposed approach can effectively address the two problems concerned above. The proposed approach is compared with some representative approaches, including ABOD, KNN, LOF, SOS and SOD, on eight well-known public datasets. The experiments show that compared with them, the proposed approach has advantages in both precision and recall rate, and can more accurately distinguish between normal data and outliers.

关键词： high-dimensional data outlier detection autoencoder Robust PCA

来源：评论

学校读者我要写书评

暂无评论

A dual linear autoencoder approach for vessel trajectory prediction using historical AIS data

引用

OCEAN ENGINEERING 2020年 209卷 107478-107478页

作者： Murray, Brian Perera, Lokukaluge Prasad UiT Arctic Univ Norway Tromso Norway

Advances in artificial intelligence are driving the development of intelligent transportation systems, with the purpose of enhancing the safety and efficiency of such systems. One of the most important aspects of maritime safety is effective collision avoidance. In this study, a novel dual linear autoencoder approach is suggested to predict the future trajectory of a selected vessel. Such predictions can serve as a decision support tool to evaluate the future risk of ship collisions. Inspired by generative models, the method suggests to predict the future trajectory of a vessel based on historical AIS data. Using unsupervised learning to facilitate trajectory clustering and classification, the method utilizes a cluster of historical AIS trajectories to predict the trajectory of a selected vessel. Similar methods predict future states iteratively, where states are dependent upon the prior predictions. The method in this study, however, suggests predicting an entire trajectory, where all states are predicted jointly. Further, the method estimates a latent distribution of the possible future trajectories of the selected vessel. By sampling from this distribution, multiple trajectories are predicted. The uncertainties of the predicted vessel positions are also quantified in this study.

关键词： Maritime situation awareness Trajectory prediction Collision avoidance Machine learning autoencoder AIS

来源：评论

学校读者我要写书评

暂无评论

5G Uplink Frequency Offset Compensation with an autoencoder

5G Uplink Frequency Offset Compensation with an Autoencoder

引用

作者： Pitkänen, Mikko Aalto University

学位级别：硕士

The emerging fifth generation 5G cellular networks face strict requirements for the quality of service. Accurate and fast algorithms need to be deployed to deliver the promised high data rates and low latencies. One of those algorithms, the frequency offset compensation, is the focal point of this thesis. The goal is to develop a machine learning model for frequency offset compensation in a 5G base station receiver. The feasibility of the model is evaluated based on comparisons to an existing conventional algorithm. Denoising autoencoders are selected as the machine learning method for this thesis. The most significant novelty of this approach is that an autoencoder does not use an external estimate of the frequency offset. The autoencoder models are developed with the PyTorch framework in the Python programming language. The development work is conducted in three paths, each focusing on one autoencoder architecture. The data for the machine learning solution is generated with a link-level simulator that can simulate communication scenarios between a transmitter and a receiver. In addition to monitoring the autoencoders' performance at training time, the trained models are integrated into the link-level simulator to examine their impact on the whole signal processing chain of the receiver. Most of the developed models demonstrate poorer performance than the conventional algorithm. It was verified that the training data contains all the needed information in order to compensate for the frequency offset ideally, but the autoencoders did not learn to utilize this information. However, the positive results in this thesis indicate that an autoencoder-based algorithm can outperform the conventional method especially at higher frequency offsets. It was also discovered that an autoencoder trained for a narrow frequency offset range is able to generalize outside of that range to some extent.

关键词： carrier frequency offset frequency offset compensation OFDM 5G NR machine learning autoencoder

来源：评论

学校读者我要写书评

暂无评论

Selective Unsupervised Learning-Based Wi-Fi Fingerprint System Using autoencoder and GAN

引用

IEEE INTERNET OF THINGS JOURNAL 2020年第3期7卷 1898-1909页

作者： Seong, J. H. Seo, D. H. Korea Maritime & Ocean Univ Adv IT & Ship Convergence Ctr Busan 49112 South Korea Korea Maritime & Ocean Univ Div Elect & Elect Informat Engn Busan 49112 South Korea

In this article, we propose an automatic Wi-Fi fingerprint system that combines an unsupervised dual radio mapping (UDRM) algorithm with the aim of reducing the time-cost needed to acquire Wi-Fi signals. Our proposed system is appropriate for indoor environments and utilizes a minimum description length principle (MDLP)-based radio map feedback (RMF) algorithm that simultaneously optimizes and updates the radio map. In the training phase, the proposed UDRM algorithm generates a radio map of the entire building based on the measured radio map of one reference floor. It does this by selectively applying a modified autoencoder and a generative adversarial network according to the spatial structures. Our proposed learning-based UDRM algorithm does not require labeled data, which is essential for supervised and semisupervised learning algorithms. It has a relatively low dependence on received signal strength indicator (RSSI) data sets. Our proposed RMF algorithm analyzes the distribution characteristics of the RSSIs for newly measured access points (APs) and feeds the analyzed results back to the radio map. The MDLP, which is applied to the proposed algorithm, improves the positioning performance and optimizes the size of the radio map by preventing the indefinite updating of the RSSI and by updating the newly added APs in the radio map.

关键词： Fingerprint recognition Sea measurements Generators Wireless fidelity Generative adversarial networks Unsupervised learning Buildings autoencoder fingerprint generative adversarial network (GAN) radio map unsupervised learning

来源：评论

学校读者我要写书评

暂无评论

Modified autoencoder Training and Scoring for Robust Unsupervised Anomaly Detection in Deep Learning

引用

IEEE ACCESS 2020年 8卷 101824-101833页

作者： Merrill, Nicholas Eskandarian, Azim Virginia Tech Dept Mech Engn Blacksburg VA 24060 USA

The autoencoder (AE) is a fundamental deep learning approach to anomaly detection. AEs are trained on the assumption that abnormal inputs will produce higher reconstruction errors than normal ones. In practice, however, this assumption is unreliable in the unsupervised case, where the training data may contain anomalous examples. Given sufficient capacity and training time, an AE can generalize to such an extent that it reliably reconstructs anomalies. Consequently, the ability to distinguish anomalies via reconstruction errors is diminished. We respond to this limitation by introducing three new methods to more reliably train AEs for unsupervised anomaly detection: cumulative error scoring (CES), percentile loss (PL), and early stopping via knee detection. We demonstrate significant improvements over conventional AE training on image, remote-sensing, and cybersecurity datasets.

关键词： Training Anomaly detection Image reconstruction Machine learning Training data Reliability Data models Hyperspectral HSI deep learning anomaly detection unsupervised autoencoder outlier detection condition monitoring

来源：评论

学校读者我要写书评

暂无评论

Supervised feature learning by adversarial autoencoder approach for object classification in dual X-ray image of luggage

引用

JOURNAL OF INTELLIGENT MANUFACTURING 2020年第5期31卷 1101-1112页

作者： Chouai, Mohamed Merah, Mostefa Sancho-Gomez, Jose-Luis Mimi, Malika Mostaganem Univ Dept Elect Engn Signals & Syst Lab Site 1Route Belahcel Mostaganem 27000 Algeria Univ Politecn Cartagena Informat & Commun Technol Dept TDAM Res Grp Data Proc & Machine Learning Plaza Hosp 1 Murcia 30202 Spain

X-ray inspection by control officers is not always consistent when inspecting baggage since this task are monotonous, tedious and tiring for human inspectors. Thus, a semi-automatic inspection makes sense as a solution in this case. In this perspective, the study presents a novel feature learning model for object classification in luggage dual X-ray images in order to detect explosives objects and firearms. We propose to use supervised feature learning by autoencoders approach. Object detection is performed by a modified YOLOv3 to detect all the presented objects without classification. The features learning is carried out by labeled adversarial autoencoders. The classification is performed by a support vector machine to classify a new object as explosive, firearms or non-threatening objects. To show the superiority of our proposed system, a comparative analysis was carried out to several methods of deep learning. The results indicate that the proposed system leads to efficient objects classification in complex environments, achieving an accuracy of 98.00% and 96.50% in detection of firearms and explosive objects respectively.

关键词： Transport security Baggage control Object detection YOLOv3 Features learning autoencoder

来源：评论

学校读者我要写书评

暂无评论

Deep Manifold Preserving autoencoder for Classifying Breast Cancer Histopathological Images

引用

IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS 2020年第1期17卷 91-101页

作者： Feng, Yangqin Zhang, Lei Mo, Juan Sichuan Univ Machine Intelligence Lab Coll Comp Sci Chengdu 610065 Sichuan Peoples R China

Classifying breast cancer histopathological images automatically is an important task in computer assisted pathology analysis. However, extracting informative and non-redundant features for histopathological image classification is challenging due to the appearance variability caused by the heterogeneity of the disease, the tissue preparation, and staining processes. In this paper, we propose a new feature extractor, called deep manifold preserving autoencoder, to learn discriminative features from unlabeled data. Then, we integrate the proposed feature extractor with a softmax classifier to classify breast cancer histopathology images. Specifically, it learns hierarchal features from unlabeled image patches by minimizing the distance between its input and output, and simultaneously preserving the geometric structure of the whole input data set. After the unsupervised training, we connect the encoder layers of the trained deep manifold preserving autoencoder with a softmax classifier to construct a cascade model and fine-tune this deep neural network with labeled training data. The proposed method learns discriminative features by preserving the structure of the input datasets from the manifold learning view and minimizing reconstruction error from the deep learning view from a large amount of unlabeled data. Extensive experiments on the public breast cancer dataset (BreaKHis) demonstrate the effectiveness of the proposed method.

关键词： Breast cancer Manifolds Feature extraction Neural networks Training Computer architecture Histopathological image classification breast cancer diagnose manifold learning autoencoder deep neural networks

来源：评论

学校读者我要写书评

暂无评论

A Fast Deep autoencoder for high-dimensional and sparse matrices in recommender systems

引用

NEUROCOMPUTING 2020年 412卷 381-391页

作者： Jiang, Jiajia Li, Weiling Dong, Ani Gou, Quanhui Luo, Xin Dongguan Univ Technol Sch Comp Sci & Technol Dongguan 523808 Guangdong Peoples R China Dongguan Univ Technol Dept Comp & Informat Sci City Coll Dongguan 523419 Guangdong Peoples R China Lenovo US Morrisville NC 27560 USA CloudWalk Dept Big Data Anal Tech Hengrui Chongqing Artificial Intelligence Res Ctr Chongqing 401331 Peoples R China

A latent factor analysis (LFA)-based model has outstanding performance in extracting desired patterns from High-dimensional and Sparse (HiDS) data for building a recommender systems. However, they mostly fail in acquiring non-linear features from an HiDS matrix. An autoencoder (AE)-based model can address this issue efficiently, but it requires filling unknown data of an HiDS matrix with pre assumptions that leads to the following limitations: a) prefilling unknown data of an HiDS matrix might skew its known data distribution to generate ridiculous recommendations;and b) incorporating a deep AE-style structure to improve its representative learning ability. Experimental results on three HiDS matrices from real recommender systems show that an FDAE-based model significantly outperforms state-of-the-art recommenders in terms of recommendation accuracy. Meanwhile, its computational efficiency is comparable with the most efficient recommenders with the help of parallelization. (c) 2020 Elsevier B.V. All rights reserved.

关键词： Recommender System Collaborative Filtering Latent Factor Analysis autoencoder Parallelization High-dimensional and Sparse Matrix Deep Neural Network

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：