检索结果-内蒙古大学图书馆

Report on the AAPM deep-learning sparse-view CT (DL-sparse-view CT) Grand Challenge

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Sidky, Emil Y. Pan, Xiaochuan Department of Radiology The University of Chicago 5841 S. Maryland Ave. ChicagoIL60637 United States

Purpose: The purpose of the challenge is to find the deep-learning technique for sparse-view CT image reconstruction that can yield the minimum RMSE under ideal conditions, thereby addressing the question of whether or not deep learning can solve inverse problems in imaging. Methods: The challenge set-up involves a 2D breast CT simulation, where the simulated breast phantom has random fibro-glandular structure and high-contrast specks. The phantom allows for arbitrarily large training sets to be generated with perfectly known truth. The training set consists of 4000 cases where each case consists of the truth image, 128-view sinogram data, and the corresponding 128-view filtered back-projection (FBP) image. The networks are trained to predict the truth image from either the sinogram or FBP data. Geometry information is not provided. The participating algorithms are tested on a data set consisting of 100 new cases. Results: About 60 groups participated in the challenge at the validation phase, and 25 groups submitted test-phase results along with reports on their deep-learning methodology. The winning team improved reconstruction accuracy by two orders of magnitude over our previous CNN-based study on a similar test problem. Conclusions: The DL-sparse-view challenge provides a unique opportunity to examine the state-of-the-art in deep-learning techniques for solving the sparse-view CT inverse problem. © 2021, CC BY.

关键词： learning algorithms

Policy Impact Analysis of Hierarchical Shift of Travellers’ Public Transport Dependence Under the Pandemic Condition Using Improved Apriori Algorithm

学校读者我要写书评

暂无评论

SSRN

SSRN 2022年

作者： Hu, Song Wang, Maolin Zhang, Mengyuan Weng, Jiancheng Zhou, Wei Beijing Key Laboratory of Traffic Engineering Beijing University of Technology Beijing100124 China Ministry of Transport of the People’s Republic of China Beijing100736 China

Exploring travellers’ dependence on public transportation (PT) is conducive to understanding individuals’ or groups’ travel choice behaviour and optimizing PT operation organizations. To explore the internal causal relationship between travellers’ dependence on PT and the key influencing factors under the pandemic condition , an online travel survey is designed and implemented in Beijing, China. The individual PT travel chains and travel knowledge graphs are constructed by associating and matching the multisource PT big data and travel survey data. To analyse the heterogeneity of travel behaviour characteristics of different groups, the K -means algorithm is used to identify and classify travellers’ PT dependence levels. Then, an improved Apriori algorithm is developed to mine the frequent association rules of groups under the Corona Virus Disease 2019 (COVID-19 ) epidemic condition. Then, the policy implications of PT dependence hierarchy transfer are developed based on the differences between indicators of association rules. The results show that the travellers are divided into four clusters based on PT dependency levels by clustering the behavioural features. The association rules of travellers’ PT dependence are significantly different among different clusters. The lower the PT dependence level is, the lower the co-occurrence degree and occurrence probability of association rules are. Furthermore, the total distance from origin and destination to PT transit and car availability are the key indicators for each cluster to enhance their PT dependence levels, while whether the routes within high-risk epidemic areas and the support degree of relatives and friends for PT usage are important factors for improving the PT usage behaviour of the clusters with the relatively high dependence level during the epidemic period. The discovered frequent patterns and association rules describe the relationships between key influencing indicators and travellers’ PT dependence. Fin

关键词： learning algorithms

CPL Criterion Functions and learning algorithms Linked to the Linear Separability Concept

学校读者我要写书评

暂无评论

<i>CPL</i> Criterion Functions and Learning Algorithms Linke...

14th International Conference on Engineering Applications of Neural Networks (EANN)

作者： Bobrowski, Leon Bialystok Tech Univ Fac Comp Sci Bialystok Poland

ISBN: (纸本)9783642410123;9783642410130

Linear separabilty of learning sets is a basic concept of neural networks theory. Exploration of the linear separability can be based on the minimization of the perceptron criterion function. Modification of the perceptron criterion function have been proposed recently aimed at feature selection problem. The modified criterion functions allows, among others, for discovering minimal feature subset that assure linear separability. learning algorithm linked to the modified function is formulated in the paper.

关键词： learning sets linear separability perceptron criterion function modified criterion function learning algorithms

Model-contrastive federated learning

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Li, Qinbin He, Bingsheng Song, Dawn National University of Singapore Singapore Uc Berkeley United States

Federated learning enables multiple parties to collaboratively train a machine learning model without communicating their local data. A key challenge in federated learning is to handle the heterogeneity of local data distribution across parties. Although many studies have been proposed to address this challenge, we find that they fail to achieve high performance in image datasets with deep learning models. In this paper, we propose MOON: Modelcontrastive federated learning. MOON is a simple and effective federated learning framework. The key idea of MOON is to utilize the similarity between model representations to correct the local training of individual parties, i.e., conducting contrastive learning in model-level. Our extensive experiments show that MOON significantly outperforms the other state-of-the-art federated learning algorithms on various image classification tasks. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms

Discriminative learning for probabilistic context-free grammars based on generalized H-criterion

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Maca, Mauricio Benedí, José Miguel Sánchez, Joan Andreu Department of Mathematics Universidad del Cauca Colombia PRHLT Research Center Universitat Politécnica de Valéncia Spain

We present a formal framework for the development of a family of discriminative learning algorithms for Probabilistic Context-Free Grammars (PCFGs) based on a generalization of criterion-H. First of all, we propose the H-criterion as the objective function and the Growth Transformations as the optimization method, which allows us to develop the final expressions for the estimation of the parameters of the PCFGs. And second, we generalize the H-criterion to take into account the set of reference interpretations and the set of competing interpretations, and we propose a new family of objective functions that allow us to develop the expressions of the estimation transformations for PCFGs. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms

A loss curvature perspective on training instability in deep learning

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Gilmer, Justin Ghorbani, Behrooz Garg, Ankush Kudugunta, Sneha Neyshabur, Behnam Cardoze, David Dahl, George Nado, Zack Firat, Orhan

In this work, we study the evolution of the loss Hessian across many classification tasks in order to understand the effect the curvature of the loss has on the training dynamics. Whereas prior work has focused on how different learning rates affect the loss Hessian observed during training, we also analyze the effects of model initialization, architectural choices, and common training heuristics such as gradient clipping and learning rate warmup. Our results demonstrate that successful model and hyperparameter choices allow the early optimization trajectory to either avoid- or navigate out of-regions of high curvature and into flatter regions that tolerate a higher learning rate. Our results suggest a unifying perspective on how disparate mitigation strategies for training instability ultimately address the same underlying failure mode of neural network optimization, namely poor conditioning. Inspired by the conditioning perspective, we show that learning rate warmup can improve training stability just as much as batch normalization, layer normalization, MetaInit, GradInit, and Fixup initialization. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms

Fractional-order learning algorithm for PID neural network decoupling control based on sparrow search algorithm

学校读者我要写书评

暂无评论

Research Square

Research Square 2021年

作者： Wu, Min Ding, Jie Yuan, Tingting Xiao, Min Nanjing University of Posts and Telecommunications Jangsu China

Research on control of multi-variable system with strong coupling has been a significant issue in industry. To accurately eliminate the coupling between system variables and improve the control effect, decoupling control techniques are investigated. In this paper, a decoupling control scheme based on fractional-order proportion integration differentiation neural network and sparrow search algorithm (SSA-FPIDNN) is proposed, where sparrow search algorithm is employed to derive the optimal initial weights, preventing the weights from falling into the local optimum, while the fractional-order algorithm is used to correct its connection weights to improve control accuracy. Compared with traditional PIDNN, the proposed SSA-FPIDNN has better decoupling control performance, and the tracking time can be reduced significantly. Numerical simulation and engineering examples verified its effectiveness. © 2021, CC BY.

关键词： learning algorithms

Average-Reward learning and Planning with Options

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Wan, Yi Naik, Abhishek Sutton, Richard S. University of Alberta Amii Edmonton Canada DeepMind Edmonton Canada

We extend the options framework for temporal abstraction in reinforcement learning from discounted Markov decision processes (MDPs) to average-reward MDPs. Our contributions include general convergent off-policy inter-option learning algorithms, intra-option algorithms for learning values and models, as well as sample-based planning variants of our learning algorithms. Our algorithms and convergence proofs extend those recently developed by Wan, Naik, and Sutton. We also extend the notion of option-interrupting behavior from the discounted to the average-reward formulation. We show the efficacy of the proposed algorithms with experiments on a continuing version of the Four-Room domain. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms

Closed-loop error learning control for uncertain nonlinear systems with experimental validation on a mobile robot

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Kayacan, Erkan School of Mechanical and Mining Engineering The University of Queensland BrisbaneQLD4072 Australia

This paper develops a Closed-Loop Error learning Control (CLELC) algorithm for feedback linearizable systems with experimental validation on a mobile robot. Traditional feedback and feedforward controllers are designed based on the nominal model by using Feedback Linearization Control (FLC) method. Then, an intelligent controller is designed based on sliding mode learning algorithm that utilizes closed-loop error dynamics to learn the system behavior. The controllers are working in parallel, and the intelligent controller can gradually replace the feedback controller from the control of the system. In addition to the stability of the sliding mode learning algorithm, the closed-loop stability of an nth order feedback linearizable system is proven. The simulation results demonstrate that CLELC algorithm can improve control performance (e.g., smaller rise time, settling time and overshoot) in the absence of uncertainties, and also provides robust control performance in the presence of uncertainties as compared to traditional FLC method. To test the efficiency and efficacy of CLELC algorithm, the trajectory tracking problem of a tracked mobile robot is studied in real-time. The experimental results demonstrate that CLELC algorithm ensures high-accurate trajectory tracking performance than traditional FLC method. © 2021, CC BY.

关键词： learning algorithms