Adjustment learning and relevant component analysis

Noam Shental, Tomer Hertz, Daphna Weinshall, Misha Pavel

פרסום מחקרי: פרק בספר / בדוח / בכנספרסום בספר כנסביקורת עמיתים

תקציר

We propose a new learning approach for image retrieval, which we call adjustment learning, and demonstrate its use for face recognition and color matching. Our approach is motivated by a frequently encountered problem, namely, that variability in the original data representation which is not relevant to the task may interfere with retrieval and make it very difficult. Our key observation is that in real applications of image retrieval, data sometimes comes in small chunks - small subsets of images that come from the same (but unknown) class. This is the case, for example, when a query is presented via a short video clip. We call these groups chunklets, and we call the paradigm which uses chunklets for unsupervised learning adjustment learning. Within this paradigm we propose a linear scheme, which we call Relevant Component Analysis; this scheme uses the information in such chunklets to reduce irrelevant variability in the data while amplifying relevant variability. We provide results using our method on two problems: face recognition (using a database publicly available on the web), and visual surveillance (using our own data). In the latter application chunklets are obtained automatically from the data without the need of supervision.

שפה מקוריתאנגלית
כותר פרסום המארחComputer Vision - ECCV 2002 - 7th European Conference on Computer Vision, Proceedings
עורכיםAnders Heyden, Gunnar Sparr, Mads Nielsen, Peter Johansen
מוציא לאורSpringer Verlag
עמודים776-790
מספר עמודים15
כרך2353
מסת"ב (אלקטרוני)9783540437482
מזהי עצם דיגיטלי (DOIs)
סטטוס פרסוםפורסם - 2002
פורסם באופן חיצוניכן
אירוע7th European Conference on Computer Vision, ECCV 2002 - Copenhagen, דנמרק
משך הזמן: 28 מאי 200231 מאי 2002

סדרות פרסומים

שםLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
כרך2353
ISSN (מודפס)0302-9743
ISSN (אלקטרוני)1611-3349

כנס

כנס7th European Conference on Computer Vision, ECCV 2002
מדינה/אזורדנמרק
עירCopenhagen
תקופה28/05/0231/05/02

הערה ביבליוגרפית

Publisher Copyright:
© Springer-Verlag Berlin Heidelberg 2002.

טביעת אצבע

להלן מוצגים תחומי המחקר של הפרסום 'Adjustment learning and relevant component analysis'. יחד הם יוצרים טביעת אצבע ייחודית.

פורמט ציטוט ביבליוגרפי