The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Kaedah pengelasan berbilang label yang melampau telah digunakan secara meluas dalam tugas pengelasan skala Web seperti penandaan halaman Web dan pengesyoran produk. Dalam makalah ini, kami membentangkan kaedah pembenaman graf baru yang dipanggil "AnnexML". Pada langkah latihan, AnnexML membina a k-graf jiran terdekat bagi vektor label dan cuba menghasilkan semula struktur graf dalam ruang benam. Ramalan dilakukan dengan cekap dengan menggunakan kaedah carian jiran terdekat anggaran yang cekap meneroka yang dipelajari k-graf jiran terdekat dalam ruang benam. Kami menjalankan penilaian ke atas beberapa set data dunia sebenar berskala besar dan membandingkan kaedah kami dengan kaedah terkini. Keputusan percubaan menunjukkan bahawa AnnexML kami boleh meningkatkan ketepatan ramalan dengan ketara, terutamanya pada set data yang mempunyai ruang label yang lebih besar. Selain itu, AnnexML menambah baik pertukaran antara masa ramalan dan ketepatan. Pada tahap ketepatan yang sama, masa ramalan AnnexML adalah sehingga 58 kali lebih cepat daripada SLEEC, kaedah berasaskan benam yang terkini.
Yukihiro TAGAMI
Yahoo Japan Corporation,Kyoto University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Salinan
Yukihiro TAGAMI, "Speeding up Extreme Multi-Label Classifier by Approximate Nearest Neighbor Search" in IEICE TRANSACTIONS on Information,
vol. E101-D, no. 11, pp. 2784-2794, November 2018, doi: 10.1587/transinf.2018EDP7096.
Abstract: Extreme multi-label classification methods have been widely used in Web-scale classification tasks such as Web page tagging and product recommendation. In this paper, we present a novel graph embedding method called “AnnexML”. At the training step, AnnexML constructs a k-nearest neighbor graph of label vectors and attempts to reproduce the graph structure in the embedding space. The prediction is efficiently performed by using an approximate nearest neighbor search method that efficiently explores the learned k-nearest neighbor graph in the embedding space. We conducted evaluations on several large-scale real-world data sets and compared our method with recent state-of-the-art methods. Experimental results show that our AnnexML can significantly improve prediction accuracy, especially on data sets that have a larger label space. In addition, AnnexML improves the trade-off between prediction time and accuracy. At the same level of accuracy, the prediction time of AnnexML was up to 58 times faster than that of SLEEC, a state-of-the-art embedding-based method.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2018EDP7096/_p
Salinan
@ARTICLE{e101-d_11_2784,
author={Yukihiro TAGAMI, },
journal={IEICE TRANSACTIONS on Information},
title={Speeding up Extreme Multi-Label Classifier by Approximate Nearest Neighbor Search},
year={2018},
volume={E101-D},
number={11},
pages={2784-2794},
abstract={Extreme multi-label classification methods have been widely used in Web-scale classification tasks such as Web page tagging and product recommendation. In this paper, we present a novel graph embedding method called “AnnexML”. At the training step, AnnexML constructs a k-nearest neighbor graph of label vectors and attempts to reproduce the graph structure in the embedding space. The prediction is efficiently performed by using an approximate nearest neighbor search method that efficiently explores the learned k-nearest neighbor graph in the embedding space. We conducted evaluations on several large-scale real-world data sets and compared our method with recent state-of-the-art methods. Experimental results show that our AnnexML can significantly improve prediction accuracy, especially on data sets that have a larger label space. In addition, AnnexML improves the trade-off between prediction time and accuracy. At the same level of accuracy, the prediction time of AnnexML was up to 58 times faster than that of SLEEC, a state-of-the-art embedding-based method.},
keywords={},
doi={10.1587/transinf.2018EDP7096},
ISSN={1745-1361},
month={November},}
Salinan
TY - JOUR
TI - Speeding up Extreme Multi-Label Classifier by Approximate Nearest Neighbor Search
T2 - IEICE TRANSACTIONS on Information
SP - 2784
EP - 2794
AU - Yukihiro TAGAMI
PY - 2018
DO - 10.1587/transinf.2018EDP7096
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E101-D
IS - 11
JA - IEICE TRANSACTIONS on Information
Y1 - November 2018
AB - Extreme multi-label classification methods have been widely used in Web-scale classification tasks such as Web page tagging and product recommendation. In this paper, we present a novel graph embedding method called “AnnexML”. At the training step, AnnexML constructs a k-nearest neighbor graph of label vectors and attempts to reproduce the graph structure in the embedding space. The prediction is efficiently performed by using an approximate nearest neighbor search method that efficiently explores the learned k-nearest neighbor graph in the embedding space. We conducted evaluations on several large-scale real-world data sets and compared our method with recent state-of-the-art methods. Experimental results show that our AnnexML can significantly improve prediction accuracy, especially on data sets that have a larger label space. In addition, AnnexML improves the trade-off between prediction time and accuracy. At the same level of accuracy, the prediction time of AnnexML was up to 58 times faster than that of SLEEC, a state-of-the-art embedding-based method.
ER -