東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Deep Learning Face Representation by...

Sun, Yi.

FindBook

Google Book

Amazon

博客來

Deep Learning Face Representation by Joint Identification-Verification.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Deep Learning Face Representation by Joint Identification-Verification./
作者:	Sun, Yi.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2015,
面頁冊數:	124 p.
附註:	Source: Dissertation Abstracts International, Volume: 78-05(E), Section: B.
Contained By:	Dissertation Abstracts International78-05B(E).
標題:	Artificial intelligence. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10297334
ISBN:	9781369410396

Deep Learning Face Representation by Joint Identification-Verification.
Sun, Yi.

Deep Learning Face Representation by Joint Identification-Verification. - Ann Arbor : ProQuest Dissertations & Theses, 2015 - 124 p.

Source: Dissertation Abstracts International, Volume: 78-05(E), Section: B.

Thesis (Ph.D.)--The Chinese University of Hong Kong (Hong Kong), 2015.

This thesis studies how to make full use of the superb learning capacity of deep neural networks to learn effective feature representations for face recognition. In particular, we propose three effective supervisory signals for deep feature learning, i.e., verification, identification, and joint identification-verification supervisory signals. Verification is the binary classification task of classifying pairs of training face images as being the same person or not. It greatly reduces the intra-personal variations in the face representation. Identification classifies each training face image into one of N different identities. By classifying a large number of identities (e.g., N ≈ 10,000) simultaneously, the last hidden layer of deep neural networks would extract features with rich identity-related (inter-personal) variations. Joint identification-verification combines the first two supervisory signals. Features learned by joint identification-verification supervisory signals exhibit both rich interpersonal variations and small intra-personal variations. With these proposed deep feature learning techniques, we greatly advances the face recognition performance on the extensively evaluated LFW benchmarks. Finally, we empirically investigate three properties of the learned deep feature representations critical for the high performance: sparsity, selectiveness, and robustness.

ISBN: 9781369410396Subjects--Topical Terms:

516317
Artificial intelligence.

Deep Learning Face Representation by Joint Identification-Verification.
LDR:02347nmm a2200301 4500 001 2122313
005 20170912094020.5
008 180830s2015 ||||||||||||||||| ||eng d
020 $a 9781369410396
035 $a (MiAaPQ)AAI10297334
035 $a AAI10297334
040 $a MiAaPQ $c MiAaPQ
100 1 $a Sun, Yi. $3 1305494
245 1 0 $a Deep Learning Face Representation by Joint Identification-Verification.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2015
300 $a 124 p.
500 $a Source: Dissertation Abstracts International, Volume: 78-05(E), Section: B.
500 $a Adviser: Jianwei Huang.
502 $a Thesis (Ph.D.)--The Chinese University of Hong Kong (Hong Kong), 2015.
520 $a This thesis studies how to make full use of the superb learning capacity of deep neural networks to learn effective feature representations for face recognition. In particular, we propose three effective supervisory signals for deep feature learning, i.e., verification, identification, and joint identification-verification supervisory signals. Verification is the binary classification task of classifying pairs of training face images as being the same person or not. It greatly reduces the intra-personal variations in the face representation. Identification classifies each training face image into one of N different identities. By classifying a large number of identities (e.g., N ≈ 10,000) simultaneously, the last hidden layer of deep neural networks would extract features with rich identity-related (inter-personal) variations. Joint identification-verification combines the first two supervisory signals. Features learned by joint identification-verification supervisory signals exhibit both rich interpersonal variations and small intra-personal variations. With these proposed deep feature learning techniques, we greatly advances the face recognition performance on the extensively evaluated LFW benchmarks. Finally, we empirically investigate three properties of the learned deep feature representations critical for the high performance: sparsity, selectiveness, and robustness.
590 $a School code: 1307.
650 4 $a Artificial intelligence. $3 516317
650 4 $a Computer engineering. $3 621879
650 4 $a Information technology. $3 532993
690 $a 0800
690 $a 0464
690 $a 0489
710 2 $a The Chinese University of Hong Kong (Hong Kong). $3 1017547
773 0 $t Dissertation Abstracts International $g 78-05B(E).
790 $a 1307
791 $a Ph.D.
792 $a 2015
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10297334