Indexed Optimization: Learning Ramp-Loss SVM in Sublinear Time
碩士 === 國立臺灣大學 === 資訊工程學研究所 === 100 === Multidimensional indexing has been frequently used for sublinear-time nearest neighbor search in various applications. In this paper, we demonstrate how this technique can be integrated into learning problem with sublinear sparsity like ramp-loss SVM. We propos...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | en_US |
Published: |
2012
|
Online Access: | http://ndltd.ncl.edu.tw/handle/73543025733430311023 |
id |
ndltd-TW-100NTU05392071 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-100NTU053920712015-10-13T21:50:17Z http://ndltd.ncl.edu.tw/handle/73543025733430311023 Indexed Optimization: Learning Ramp-Loss SVM in Sublinear Time 利用索引技術達到低於線性時間的支持向量機訓練 EN-HSU YEN 嚴恩勗 碩士 國立臺灣大學 資訊工程學研究所 100 Multidimensional indexing has been frequently used for sublinear-time nearest neighbor search in various applications. In this paper, we demonstrate how this technique can be integrated into learning problem with sublinear sparsity like ramp-loss SVM. We propose an outlier-free convex-relaxation for ramp-loss SVM and an indexed optimization algorithm which solves large-scale problem in sublinear-time even when data cannot fit into memory. We compare our algorithm with state-of-the-art linear hinge-loss solver and ramp-loss solver in both sufficient and limited memory conditions, where our algorithm not only learns several times faster but achieves more accurate result on noisy and large-scale datasets. 林守德 2012 學位論文 ; thesis 32 en_US |
collection |
NDLTD |
language |
en_US |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立臺灣大學 === 資訊工程學研究所 === 100 === Multidimensional indexing has been frequently used for sublinear-time nearest neighbor search in various applications. In this paper, we demonstrate how this technique can be integrated into learning problem with sublinear sparsity like ramp-loss SVM. We propose an outlier-free convex-relaxation for ramp-loss SVM and an indexed optimization algorithm which solves large-scale problem in sublinear-time even when data cannot fit into memory. We compare our algorithm with state-of-the-art linear hinge-loss solver and ramp-loss solver in both sufficient and limited memory conditions, where our algorithm not only learns several times faster but achieves more accurate result on noisy and large-scale datasets.
|
author2 |
林守德 |
author_facet |
林守德 EN-HSU YEN 嚴恩勗 |
author |
EN-HSU YEN 嚴恩勗 |
spellingShingle |
EN-HSU YEN 嚴恩勗 Indexed Optimization: Learning Ramp-Loss SVM in Sublinear Time |
author_sort |
EN-HSU YEN |
title |
Indexed Optimization: Learning Ramp-Loss SVM in Sublinear Time |
title_short |
Indexed Optimization: Learning Ramp-Loss SVM in Sublinear Time |
title_full |
Indexed Optimization: Learning Ramp-Loss SVM in Sublinear Time |
title_fullStr |
Indexed Optimization: Learning Ramp-Loss SVM in Sublinear Time |
title_full_unstemmed |
Indexed Optimization: Learning Ramp-Loss SVM in Sublinear Time |
title_sort |
indexed optimization: learning ramp-loss svm in sublinear time |
publishDate |
2012 |
url |
http://ndltd.ncl.edu.tw/handle/73543025733430311023 |
work_keys_str_mv |
AT enhsuyen indexedoptimizationlearningramplosssvminsublineartime AT yánēnxù indexedoptimizationlearningramplosssvminsublineartime AT enhsuyen lìyòngsuǒyǐnjìshùdádàodīyúxiànxìngshíjiāndezhīchíxiàngliàngjīxùnliàn AT yánēnxù lìyòngsuǒyǐnjìshùdádàodīyúxiànxìngshíjiāndezhīchíxiàngliàngjīxùnliàn |
_version_ |
1718068875944263680 |