國立虎尾科技大學 |

The Application of Optimization Techniques in Machine Learning.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	The Application of Optimization Techniques in Machine Learning./
Author:	Pang, Haotian.
Description:	1 online resource (130 pages)
Notes:	Source: Dissertation Abstracts International, Volume: 79-04(E), Section: B.
Subject:	Electrical engineering. -
Online resource:	click for full text (PQDT)
ISBN:	9780355323153

The Application of Optimization Techniques in Machine Learning.
Pang, Haotian.

The Application of Optimization Techniques in Machine Learning. - 1 online resource (130 pages)

Source: Dissertation Abstracts International, Volume: 79-04(E), Section: B.

Thesis (Ph.D.)--Princeton University, 2017.

Includes bibliographical references

The past decades have witnessed significantly progress in machine learning, and solving these problems requires the advancing in optimization techniques. High dimensional sparse learning has imposed a great computational challenge to large scale data analysis. In this dissertation, parametric simplex method is applied to solve a broad class of sparse learning approaches, which can be formulated as linear programs parametrized by a regularization parameter. There are serious drawbacks to the existing methods for solving these types of problems as tuning the parameter for the desired solution is very inefficient. The customized parametric simplex method is introduced and it uses the unknown weighting factor as the parameter and provides a powerful and efficient way to address these shortcomings. Although the simplex method has an exponential complexity in the worse case, it has been shown that the parametric simplex method is an appropriate method for these cases when the expected solution is sparse. An R package named fastclime which efficiently solves a variety machine learning problems by the customized parametric simplex method is developed. A convex optimization method named Inexact Peaceman-Rachford Splitting Method (IPRSM) is studied. Similar to the Alternating Direction Method of Multiplier (ADMM), the strictly contractive Peaceman-Rachford Splitting Method (PRSM) is also used to solve a convex minimization problem with linear constraints and a separable objective function. In many applications, it is quite expensive to obtain exact solutions to these subproblems. The inexact methods intend to solve the iterative subproblems when the exact solutions do not exist or they are hard to obtain. Finally, the new graph Perceptron algorithm, a graph estimation method which performs on online binary classification problems is proposed. The new graph Perceptron algorithm is a new kernel based Perceptron derived from online class-action and extend to online graph estimation with a new kernel trick.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2018

Mode of access: World Wide Web

ISBN: 9780355323153Subjects--Topical Terms:

596380
Electrical engineering.
Index Terms--Genre/Form:

554714
Electronic books.

The Application of Optimization Techniques in Machine Learning.
LDR:03177ntm a2200325K 4500 001 912386
005 20180608141653.5
006 m o u
007 cr mn||||a|a||
008 190606s2017 xx obm 000 0 eng d
020 $a 9780355323153
035 $a (MiAaPQ)AAI10608493
035 $a (MiAaPQ)princeton:12264
035 $a AAI10608493
040 $a MiAaPQ $b eng $c MiAaPQ
100 1 $a Pang, Haotian. $3 1184729
245 1 4 $a The Application of Optimization Techniques in Machine Learning.
264 0 $c 2017
300 $a 1 online resource (130 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertation Abstracts International, Volume: 79-04(E), Section: B.
500 $a Adviser: Robert Vanderbei.
502 $a Thesis (Ph.D.)--Princeton University, 2017.
504 $a Includes bibliographical references
520 $a The past decades have witnessed significantly progress in machine learning, and solving these problems requires the advancing in optimization techniques. High dimensional sparse learning has imposed a great computational challenge to large scale data analysis. In this dissertation, parametric simplex method is applied to solve a broad class of sparse learning approaches, which can be formulated as linear programs parametrized by a regularization parameter. There are serious drawbacks to the existing methods for solving these types of problems as tuning the parameter for the desired solution is very inefficient. The customized parametric simplex method is introduced and it uses the unknown weighting factor as the parameter and provides a powerful and efficient way to address these shortcomings. Although the simplex method has an exponential complexity in the worse case, it has been shown that the parametric simplex method is an appropriate method for these cases when the expected solution is sparse. An R package named fastclime which efficiently solves a variety machine learning problems by the customized parametric simplex method is developed. A convex optimization method named Inexact Peaceman-Rachford Splitting Method (IPRSM) is studied. Similar to the Alternating Direction Method of Multiplier (ADMM), the strictly contractive Peaceman-Rachford Splitting Method (PRSM) is also used to solve a convex minimization problem with linear constraints and a separable objective function. In many applications, it is quite expensive to obtain exact solutions to these subproblems. The inexact methods intend to solve the iterative subproblems when the exact solutions do not exist or they are hard to obtain. Finally, the new graph Perceptron algorithm, a graph estimation method which performs on online binary classification problems is proposed. The new graph Perceptron algorithm is a new kernel based Perceptron derived from online class-action and extend to online graph estimation with a new kernel trick.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2018
538 $a Mode of access: World Wide Web
650 4 $a Electrical engineering. $3 596380
650 4 $a Operations research. $3 573517
655 7 $a Electronic books. $2 local $3 554714
690 $a 0544
690 $a 0796
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a Princeton University. $b Electrical Engineering. $3 1181685
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10608493 $z click for full text (PQDT)