請用此 Handle URI 來引用此文件：
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/48971完整後設資料紀錄
| DC 欄位 | 值 | 語言 |
|---|---|---|
| dc.contributor.advisor | 林智仁 | |
| dc.contributor.author | Ya Zhu | en |
| dc.contributor.author | 朱雅 | zh_TW |
| dc.date.accessioned | 2021-06-15T11:12:40Z | - |
| dc.date.available | 2021-11-22 | |
| dc.date.copyright | 2016-11-22 | |
| dc.date.issued | 2016 | |
| dc.date.submitted | 2016-08-22 | |
| dc.identifier.citation | BIBLIOGRAPHY
[1] K.-W. Chang, C.-J. Hsieh, and C.-J. Lin. Coordinate descent method for large-scale L2-loss linear SVM. Journal of Machine Learning Research, 9:1369{1398, 2008. [2] A. R. Conn, N. I. M. Gould, and P. L. Toint. Trust-region Methods. Society for Industrial and Applied Mathematics, Philadelphia, 2000. [3] R.-E. Fan, K.-W. Chang, C.-J. Hsieh, X.-R. Wang, and C.-J. Lin. LIBLINEAR: a library for large linear classi cation. Journal of Machine Learning Research, 9:1871{1874, 2008. [4] S. S. Keerthi and D. DeCoste. A modi ed nite Newton method for fast solution of large scale linear SVMs. Journal of Machine Learning Research, 6:341{361, 2005. [5] C.-J. Lin and J. J. Mor e. Newton's method for large-scale bound constrained problems. SIAM Journal on Optimization, 9:1100{1127, 1999. [6] C.-J. Lin, R. C. Weng, and S. S. Keerthi. Trust region Newton method for large-scale logistic regression. Journal of Machine Learning Research, 9:627{650, 2008. [7] O. L. Mangasarian. A nite Newton method for classi cation. Optimization Methods and Software, 17(5):913{929, 2002. [8] T. Steihaug. The conjugate gradient method and trust regions in large scale optimization. SIAM Journal on Numerical Analysis, 20(3):626{637, 1983. [9] P.-W. Wang, C.-P. Lee, and C.-J. Lin. The common directions method for regularized loss minimization. Technical report, National Taiwan University, 2016. | |
| dc.identifier.uri | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/48971 | - |
| dc.description.abstract | 線性分類的一個主要任務是解決最小化問題。雖然有很多最優化技術可以方便地運用，但是對於大規模資料需要有特殊的演算法改進。典型的無規則最優化演算法迭代地進行兩個過程：尋找一個好的下降方向，然後調整合適的步長。過去的線性分類發展主要集中在尋找下降方向，而很少考慮調整步長。在這篇論文中，我們提出了一種新穎的調整信賴區間大小的方法來間接地調整步長。這個設定嵌於一個針對邏輯迴歸和L2損失支援向量機的信賴區間牛頓法中。實驗證明我們的新方法極大地領先現有的大規模線性分類方法。 | zh_TW |
| dc.description.abstract | The main task in training a linear classification is to solve an unconstrained minimization problem. While many optimization techniques can be conveniently applied, special algorithmic development and implementation tricks are needed to handle large scale data sets. Typically, an unconstrained optimization method iteratively conduct two procedures of finding a good direction and then deciding a suitable step size. Past developments for linear classification focus on finding the direction, but little attention has been paid on adjusting the step size. In this work, we develop novel techniques to adjust the trust-region size that indirectly decides the step length. The setting is embedded into a trust region Newton method for logistic regression and L2-loss SVM. Experiments indicate that our new settings significantly outperform existing implementations for large-scale linear classification. | en |
| dc.description.provenance | Made available in DSpace on 2021-06-15T11:12:40Z (GMT). No. of bitstreams: 1 ntu-105-R02944047-1.pdf: 5049855 bytes, checksum: 6f094ac54a550eabf60aba475eef7a6d (MD5) Previous issue date: 2016 | en |
| dc.description.tableofcontents | TABLE OF CONTENTS
口試委員會審定書: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : i 中文摘要: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : ii ABSTRACT : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : iii LIST OF FIGURES : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : vi LIST OF TABLES : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : viii CHAPTER I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 II. Truncated Newton Methods and The Decision of Step Length 4 2.1 Line Search Methods . . . . . . . . . . . . . . . . . . . . . . . . 7 2.2 Trust Region Methods . . . . . . . . . . . . . . . . . . . . . . . 8 2.2.1 A Simple Rule for Updating the Trust Region . . . . 10 2.2.2 A Rule to Update Trust Region by Incorporating Ad- ditional Information . . . . . . . . . . . . . . . . . . . 10 2.3 Some Simple Comparisons . . . . . . . . . . . . . . . . . . . . . 11 III. Analysis and New Update Rules . . . . . . . . . . . . . . . . . . . 14 3.1 Analysis of Situations where Inappropriate Step Lengths are Obtained . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3.2 Accurate Minimization of f(w^k + as^k) . . . . . . . . . . . . . . 17 3.2.1 Newton Method . . . . . . . . . . . . . . . . . . . . . 19 3.2.2 Bisection Method . . . . . . . . . . . . . . . . . . . . 19 3.2.3 Cubic Interpolation . . . . . . . . . . . . . . . . . . . 20 3.3 Our New Update Rules . . . . . . . . . . . . . . . . . . . . . . 22 IV. Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 4.1 Data Sets and Experimental Settings . . . . . . . . . . . . . . . 23 4.2 Effectiveness of the Proposed Techniques in Chapter III . . . . 25 4.3 Effectiveness of the Proposed Update Rule . . . . . . . . . . . . 27 V. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 APPENDICES : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 31 BIBLIOGRAPHY : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 56 | |
| dc.language.iso | en | |
| dc.subject | 信賴區間 | zh_TW |
| dc.subject | 線性搜尋 | zh_TW |
| dc.subject | 牛頓法 | zh_TW |
| dc.subject | 大規模線性分類 | zh_TW |
| dc.subject | trust region | en |
| dc.subject | large-scale linear classification | en |
| dc.subject | line search | en |
| dc.subject | Newton method | en |
| dc.title | 一種基於新的信賴區間更新規則之高效牛頓法於大規模線性分類 | zh_TW |
| dc.title | An Efficient Newton Method with Novel Trust Region Update Rules for Large-scale Linear Classification | en |
| dc.type | Thesis | |
| dc.date.schoolyear | 104-2 | |
| dc.description.degree | 碩士 | |
| dc.contributor.oralexamcommittee | 林軒田,李育杰 | |
| dc.subject.keyword | 大規模線性分類,信賴區間,牛頓法,線性搜尋, | zh_TW |
| dc.subject.keyword | large-scale linear classification,trust region,Newton method,line search, | en |
| dc.relation.page | 56 | |
| dc.identifier.doi | 10.6342/NTU201603529 | |
| dc.rights.note | 有償授權 | |
| dc.date.accepted | 2016-08-22 | |
| dc.contributor.author-college | 電機資訊學院 | zh_TW |
| dc.contributor.author-dept | 資訊網路與多媒體研究所 | zh_TW |
| 顯示於系所單位： | 資訊網路與多媒體研究所 | |
文件中的檔案：
| 檔案 | 大小 | 格式 | |
|---|---|---|---|
| ntu-105-1.pdf 未授權公開取用 | 4.93 MB | Adobe PDF |
系統中的文件，除了特別指名其著作權條款之外，均受到著作權保護，並且保留所有的權利。