检索结果-内蒙古大学图书馆

Quasi-potential as an implicit regularizer for the loss function in the stochastic gradient descent.

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Hu, Wenqing Zhu, Zhanxing Xiong, Haoyi Huan, Jun Department of Mathematics and Statistics Missouri University of Science and Technology University of Missouri Rolla Peking University Beijing Institute of Big Data Research Beijing China Big Data Lab Baidu Inc. National Engineering Laboratory of Deep Learning Application and Technology

We interpret the variational inference of the Stochastic Gradient Descent (SGD) as minimizing a new potential function named the quasi-potential. We analytically construct the quasi-potential function in the case when the loss function is convex and admits only one global minimum point. We show in this case that the quasi-potential function is related to the noise covariance structure of SGD via a partial differential equation of Hamilton-Jacobi type. This relation helps us to show that anisotropic noise leads to faster escape than isotropic noise. We then consider the dynamics of SGD in the case when the loss function is non-convex and admits several different local minima. In this case, we demonstrate an example that shows how the noise covariance structure plays a role in "implicit regularization", a phenomenon in which SGD favors some particular local minimum points. This is done through the relation between the noise covariance structure and the quasi-potential function. Our analysis is based on Large Deviations Theory (LDT), and they are validated by numerical experiments. Copyright © 2019, The Authors. All rights reserved.

关键词： Stochastic systems

Micro-supervised disturbance learning: A perspective of representation probability distribution

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Chu, Jielei Liu, Jing Wang, Hongjun Meng, Hua Gong, Zhiguo Li, Tianrui The School of Computing and Artificial Intelligence Southwest Jiaotong University Chengdu611756 China National Engineering Laboratory of Integrated Transportation Big Data Application Technology Southwest Jiaotong University Chengdu611756 China The School of Business Sichuan University Sichuan Chengdu610065 China The State Key Laboratory of Internet of Things for Smart City Department of Computer and Information Science University of Macau China

The instability is shown in the existing methods of representation learning based on Euclidean distance under a broad set of conditions. Furthermore, the scarcity and high cost of labels prompt us to explore more expressive representation learning methods which depends on the labels as few as possible. To address these issues, the small-perturbation ideology is firstly introduced on the representation learning model based on the representation probability distribution. The positive small-perturbation information (SPI) which only depend on two labels of each cluster is used to stimulate the representation probability distribution and then two variant models are proposed to fine-tune the expected representation distribution of RBM, namely, Micro-supervised Disturbance GRBM (Micro-DGRBM) and Micro-supervised Disturbance RBM (Micro-DRBM) models. The Kullback-Leibler (KL) divergence of SPI is minimized in the same cluster to promote the representation probability distributions to become more similar in Contrastive Divergence (CD) learning. In contrast, the KL divergence of SPI is maximized in the different clusters to enforce the representation probability distributions to become more dissimilar in CD learning. To explore the representation learning capability under the continuous stimulation of the SPI, we present a deep Micro-supervised Disturbance Learning (Micro-DL) framework based on the Micro-DGRBM and Micro-DRBM models and compare it with a similar deep structure which has not any external stimulation. Experimental results demonstrate that the proposed deep Micro-DL architecture shows better performance in comparison to the baseline method, the most related shallow models and deep frameworks for clustering. © 2020, CC0.

关键词： Probability distributions

Wheat yellow rust monitoring method based on Sentinel-2 image and BPNN model

学校读者我要写书评

暂无评论

Nongye Gongcheng Xuebao/Transactions of the Chinese Society of Agricultural engineering 2019年第17期35卷 178-185页

作者： Huang, Linsheng Jiang, Jing Huang, Wenjiang Ye, Huichun Zhao, Jinling Ma, Huiqin Ruan, Chao National Engineering Research Center for Agro-Ecological Big Data Analysis & Application Anhui University 230601 China Sanya Institute of Remote Sensing Sanya572029 China Key Laboratory of Digital Earth Science Institute of Remote Sensing and Digital Earth Chinese Academy of Sciences Beijing100094 China

Wheat yellow rust is a deadly disease of winter wheat and its timely and accurate detection at regional scale is critical to ameliorate infectious yield loss and safeguard wheat production. With the development in remote sensing technology, satellite imagery with high spatial resolution and high revisiting frequency has become increasingly accessible. Remote sensing data has a unique advantage over traditional method in detecting crop diseases and controlling their spreading, including simple operation, real-time detection, high spatiotemporal resolution and targeting specific-disease, especially the multispectral satellite imagery which covers a wide range of wave bands and provides rich information related to crop diseases at regional scale. Compared to conventional broad band satellite imagery, the Sentinel-2 is a sensor with three wave bands within the edge of the red light which are sensitive to crop diseases. In this study, a Sentinel-2 image acquired in May 12, 2018 was used to extract 26 characteristic variables related to wheat yellow rust, including 3 visible bands (blue, green and red) reflectance variables, one near infrared band, 3 red-edge bands, 14 broad-bands and 5 red-edge vegetation indices. An approach combining K-means and ReliefF algorithm is proposed to filter these features. We first used the RelieF algorithm to calculate the weight of each feature and filter out 10 features most relevant to the disease. The feature with maximum weight was then taken as the initial center of the K-Means algorithm, and other features were added to form a cluster in descending order of their weight. The combination of features with the highest clustering accuracy was taken as the final input variable to the model. The optimal features, including enhanced vegetation index (EVI), structure intensive pigment index (SIPI), simple ratio index (SR), normalized red-edge2 index(NREDI2), normalized red-edge3 index (NREDI3), three wide-band vegetation indices and 2 red ed

关键词： Remote sensing

Fairness and accuracy in federated learning

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Huang, Wei Li, Tianrui Wang, Dexian Du, Shengdong Zhang, Junbo School of Information Science and Technology Southwest Jiaotong University Chengdu611756 China Institute of Artificial Intelligence Southwest Jiaotong University Chengdu China National Engineering Laboratory of Integrated Transportation Big Data Application Technology Southwest Jiaotong University Chengdu China JD Intelligent Cities Business Unit JD Digits Beijing China JD Intelligent Cities Research Beijing China

In the federated learning setting, multiple clients jointly train a model under the coordination of the central server, while the training data is kept on the client to ensure privacy. Normally, inconsistent distribution of data across different devices in a federated network and limited communication bandwidth between end devices impose both statistical heterogeneity and expensive communication as major challenges for federated learning. This paper proposes an algorithm to achieve more fairness and accuracy in federated learning (FedFa). It introduces an optimization scheme that employs a double momentum gradient, thereby accelerating the convergence rate of the model. An appropriate weight selection algorithm that combines the information quantity of training accuracy and training frequency to measure the weights is proposed. This procedure assists in addressing the issue of unfairness in federated learning due to preferences for certain clients. Our results show that the proposed FedFa algorithm outperforms the baseline algorithm in terms of accuracy and fairness. Copyright © 2020, The Authors. All rights reserved.

关键词： Machine learning

On efficient and robust metrics for RANSAC hypotheses and 3D rigid registration

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Yang, Jiaqi Huang, Zhiqiang Quan, Siwen Zhang, Qian Zhang, Yanning Cao, Zhiguo National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology School of Computer Science Northwestern Polytechnical University China School of Software Northwestern Polytechnical University Xi’an710129 China School of Electronic and Control Engineering Chang’an University Xi’an710064 China School of Resources and Environment Hubei University Wuhan430062 China National Key Laboratory of Science and Technology on Multi-spectral Information Processing School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China

This paper focuses on developing efficient and robust evaluation metrics for RANSAC hypotheses to achieve accurate 3D rigid registration. Estimating six-degree-of-freedom (6-DoF) pose from feature correspondences remains a popular approach to 3D rigid registration, where random sample consensus (RANSAC) is a de-facto choice to this problem. However, existing metrics for RANSAC hypotheses are either time-consuming or sensitive to common nuisances, parameter variations, and different application scenarios, resulting in performance deterioration in overall registration accuracy and speed. We alleviate this problem by first analyzing the contributions of inliers and outliers, and then proposing several efficient and robust metrics with different designing motivations for RANSAC hypotheses. Comparative experiments on four standard datasets with different nuisances and application scenarios verify that the proposed metrics can significantly improve the registration performance and are more robust than several state-of-the-art competitors, making them good gifts to practical applications. This work also draws an interesting conclusion, i.e., not all inliers are equal while all outliers should be equal, which may shed new light on this research problem. Copyright © 2020, The Authors. All rights reserved.

关键词： Deterioration

Preparation of regenerated cathode material lithium nickel cobalt oxide LiNi0.7 Co0.3O2 form spent lithium-ion battery

学校读者我要写书评

暂无评论

Preparation of regenerated cathode material lithium nickel c...

Chinese Materials Conference, 2018

作者： Wang, Yuehua Ma, Liwen Zhang, Yunhe Huang, Zhaojie Xi, Xiaoli College of Materials Science and Engineering Beijing University of Technology Beijing100124 China National Engineering Laboratory for Industrial Big-data Application Technology Beijing University of Technology Beijing100124 China

ISBN: (纸本)9783035713848

With the development of new energy vehicles, urgent issues have attracted considerable attention. Some power batteries have entered the scrapping period with the imperative recycling of used power batteries. Some studies have predicted that by 2020 the amount of power lithium battery scrap will reach 32.2 GWh, corresponding to ~500,000 tons, and by 2023 the scrap will reach 101 GWh, corresponding to ~1.16 million tons. In this study, nickel-cobalt-lithium LiNi0.7 Co0.3 O2 cathode materials are regenerated from spent lithium-ion battery cathode materials as the raw material, which not only aids in the reduction of pressure on the environment but also leads to the recycling of resources. First, the extraction was employed using extracting agent p204 to remove aluminum ions from an acid leaching solution. The extraction conditions for aluminum ions were: including a phase ratio of 1:2, a pH of 3, an extractant concentration of 30%, and a saponification rate of 70%. Next, the precursor was prepared by co-precipitation using sodium hydroxide and ammonia water as the precipitant and complexion agents, respectively;hence, the cathode material can be uniformly mixed at the atomic level. The precursor and lithium hydroxide were subjected to calcination at high temperature using a high-temperature solid-phase method. The Calcination conditions included an air atmosphere;a calcination temperature of 800 °C;a calcination time of 15 h, an n (precursor): n (lithium hydroxide) ratio of 1:1.1. The thermogravimetric analysis revealed that the synthesis temperature should not exceed 850 °C. X-ray diffraction analysis, scanning electron microscopy, and energy spectrum analysis of the cathode material revealed a composition comprising Li, Ni, and Co oxides. After analysis, the material obtained is lithium nickel-cobalt-oxide, LiNi0.7 Co0.3 O2, which is a positive electrode material with good crystallinity and a regular layered structure. © 2019 Trans Tech Publications, Switzerland.

关键词： Coprecipitation

Capacity analysis of ship-tugging operations in a large container port

学校读者我要写书评

暂无评论

Asian Transport Studies 2020年 6卷

作者： Kang, Liujiang Gao, Song Meng, Qiang Key Laboratory of Transport Industry of Big Data Application Technologies for Comprehensive Transport Ministry of Transport Beijing Jiaotong University 100044 China Intelligent Transportation Systems Research Center Wuhan University of Technology Wuhan 430063 China Department of Civil and Environmental Engineering National University of Singapore Singapore 117576 Singapore

In this study, we deal with the simulation problem of ship-tugging operations in a large container port. First, we build a tugboat-service network using a directed graph, where the nodes consist of tugboat bases, anchorages, and berths while the arcs represent the directed links between nodes. Second, we develop a marine traffic simulation system, which can simulate maritime traffic conditions in restricted waters, such as large container ports and busy straits. The marine traffic simulation system is a customized system. Users are allowed to set their study areas and virtual traffic environment (such as traffic demand and network structure) at will. Finally, we conduct various simulations for a real tugboat-service network of Singapore Port, which includes one anchorage, one tugboat base, two berths, and two channels between the anchorage and berths. A detailed analysis of the tugging capacity is provided based on simulation results. © 2020 The Authors

关键词： Capacity estimation Container ports Simulation system development Tugging operations

Finding robust and consistent space-time delivery paths for multi-day vehicle routing problem

学校读者我要写书评

暂无评论

Finding robust and consistent space-time delivery paths for ...

2019 IEEE Intelligent Transportation Systems Conference, ITSC 2019

作者： Zhuge, Lijuan Tong, Lu Carol Wu, Hailong Chen, Yiheng Zhou, Xuesong Beijing Jiaotong University School of Traffic and Transportation Haidian Beijing100044 China Beihang University School of Electronic and Information Engineering Beijing100091 China National Engineering Laboratory for Comprehensive Transportation Big Data Application Technology Beijing100091 China Beijing Jingdong Zhenshi Information Technology Co. Ltd. Logistics RandD Department China Arizona State University School of Sustainable Engineering and the Built Environment TempeAZ85281 United States

ISBN: (纸本)9781538670248

Facing uncertain and dynamic demands over different days, logistics companies need to adopt adaptive solutions for the last-mile delivery. Meanwhile, one of the major barriers for implementing vehicle routing problem (VRP) algorithms in real world dispatching systems is the stability of delivery driver's routes and schedules. To establish a theoretically sound and practically useful solution framework, this paper aims to optimize robust and consistent space-time paths for the multi-day VRP by providing daily schedules with limited variations from the master schedule. A multi-commodity network flow-based optimization model is proposed to minimize generalized transportation costs and the daily deviation of day-dependent space-time paths among demand analysis zones. Lagrange relaxation (LR) methods and alternating direction method of multipliers (ADMM) are used to handle complex side constraints. Experiments on illustrative networks and Beijing delivery network are developed to demonstrate the effectiveness of the proposed method. © 2019 IEEE.

关键词： Vehicle routing

Medical Image Segmentation Using Deep Learning: A Survey

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Wang, Risheng Lei, Tao Cui, Ruixia Zhang, Bingtao Meng, Hongying Nandi, Asoke K. The School of Electronic Information and Artificial Intelligence The Shaanxi Joint Laboratory of Artificial Intelligence Shaanxi University of Science and Technology Xi'An710021 China The 'Laboratory of Hepatobiliary Surgery First Affiliated Hospital' 'National Engineering Laboratory of Big Data Algorithm and Analysis Technology Research' Xi'an Jiaotong University Xi'An710049 China The School of Electronic and Information Engineering Lanzhou Jiaotong University Lanzhou730070 China The Department of Electronic and Electrical Engineering Brunel University London UxbridgeUB8 3PH United Kingdom

Deep learning has been widely used for medical image segmentation and a large number of papers has been presented recording the success of deep learning in the field. In this paper, we present a comprehensive thematic survey on medical image segmentation using deep learning techniques. This paper makes two original contributions. Firstly, compared to traditional surveys that directly divide literatures of deep learning on medical image segmentation into many groups and introduce literatures in detail for each group, we classify currently popular literatures according to a multi-level structure from coarse to fine. Secondly, this paper focuses on supervised and weakly supervised learning approaches, without including unsupervised approaches since they have been introduced in many old surveys and they are not popular currently. For supervised learning approaches, we analyze literatures in three aspects: the selection of backbone networks, the design of network blocks, and the improvement of loss functions. For weakly supervised learning approaches, we investigate literature according to data augmentation, transfer learning, and interactive segmentation, separately. Compared to existing surveys, this survey classifies the literatures very differently from before and is more convenient for readers to understand the relevant rationale and will guide them to think of appropriate improvements in medical image segmentation based on deep learning approaches. Copyright © 2020, The Authors. All rights reserved.

关键词： Surveys