机构

• 长春光学精密机械与物... [1]
• 数学与系统科学研究院 [1]
• 自动化研究所 [1]
• 沈阳自动化研究所 [1]

采集方式

• OAI收割 [4]

内容类型

• 期刊论文 [3]
• 会议论文 [1]

发表日期

• 2022 [1]
• 2020 [1]
• 2019 [1]
• 2012 [1]

学科主题

筛选

浏览/检索结果: 共4条，第1-4条 帮助 条数/页： 5101520253035404550556065707580859095100 排序方式： 请选择题名升序题名降序提交时间升序提交时间降序作者升序作者降序发表日期升序发表日期降序 Deep Neural Network Self-Distillation Exploiting Data Representation Invariance 期刊论文  OAI收割 IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2022, 卷号: 33, 期号: 1, 页码: 257-269 作者:  Xu, Ting-Bing;  Liu, Cheng-Lin   |  收藏  |  浏览/下载：31/0  |  提交时间：2022/02/16 Training  Nonlinear distortion  Data models  Neural networks  Knowledge engineering  Network architecture  Generalization error  network compression  representation invariance  self-distillation (SD)   Quantifying the generalization error in deep learning in terms of data distribution and neural network smoothness 期刊论文  OAI收割 NEURAL NETWORKS, 2020, 卷号: 130, 页码: 85-99 作者:  Jin, Pengzhan;  Lu, Lu;  Tang, Yifa;  Karniadakis, George Em   |  收藏  |  浏览/下载：31/0  |  提交时间：2021/01/14 Neural networks  Generalization error  Learnability  Data distribution  Cover complexity  Neural network smoothness   Rescaled Boosting in Classification 期刊论文  OAI收割 IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2019, 卷号: 30, 期号: 9, 页码: 2598-2610 作者:  Wang Y(王尧);  Liao, Xu;  Lin SB(林绍波)   |  收藏  |  浏览/下载：21/0  |  提交时间：2019/09/15 Boosting  generalization error  numerical convergence rate  resealed boosting (RBoosting)   Tracking error modeling of the theodolite based on GRNN method (EI CONFERENCE) 会议论文  OAI收割 2nd International Conference on Frontiers of Manufacturing and Design Science, ICFMD 2011, December 11, 2011 - December 13, 2011, Taichung, Taiwan 作者:  Li M. 收藏  |  浏览/下载：22/0  |  提交时间：2013/03/25 To meet the requirement of high tracking accuracy as well as develop more reasonable evaluation method  in this paper  the General Regression Neural Network (GRNN) has been applied to build the tracking error model of the theodolite. First  we analyze the nonlinear factors in the theodolite. Second  we discuss the principle of GRNN  including its structure  the function as well as its priors. Third  we build the tracking error model based on GRNN and verify the model through the different parameters. The result indicated that the network model based on GRNN has high accuracy and good generalization ability. It could instead the real system to a certain extent. The research in this paper has important value to the engineering practice.