Conference Paper

Multi-class Boosting for Early Classification of Sequences.

DOI: 10.5244/C.24.24 Conference: British Machine Vision Conference, BMVC 2010, Aberystwyth, UK, August 31 - September 3, 2010. Proceedings
Source: DBLP

ABSTRACT We propose a new boosting algorithm for sequence classification, in particular one that enables early classification of multiple classes. In many practical problems, we would like to classify a sequence into one of K classes as quickly as possible, without waiting for the end of the sequence. Recently, an early classification boosting algorithm was proposed for binary classification that employs a weight propagation technique. In this paper, we extend this model to a multi-class early classification. The derivation is based on the loss function approach, and the developed model is quite simple and effective. We validated the performance through experiments with real-world data, and confirmed the superiority of our approach over the previous method.

0 Bookmarks
 · 
59 Views
  • [Show abstract] [Hide abstract]
    ABSTRACT: We propose a new boosting method for classification of time sequences. In the problem of on-line classification, it is essential to classify time sequences as quickly as possible in many practical cases. This type of classification is called “early classification.” Recently, an Adaboost-based “Earlyboost” has been proposed, which is known for its efficiency. In this paper, we propose a Logitboost-based early classification for further improvements of Earlyboost. We demonstrate the structure of the proposed method, and experimentally verify its performance.
    Computer Analysis of Images and Patterns - 14th International Conference, CAIP 2011, Seville, Spain, August 29-31, 2011, Proceedings, Part I; 01/2011
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We propose a novel method for recognizing sequential patterns such as motion trajectory of biological objects (i.e., cells, organelle, protein molecules, etc.), human behavior motion, and meteorological data. In the proposed method, a local classifier is prepared for every point (or timing or frame) and then the whole pattern is recognized by majority voting of the recognition results of the local classifiers. The voting strategy has a strong benefit that even if an input pattern has a very large deviation from a prototype locally at several points, they do not severely influence the recognition result; they are treated just as several incorrect votes and thus will be neglected successfully through the majority voting. For regularizing the recognition result, we introduce partial-dependency to local classifiers. An important point is that this dependency is introduced to not only local classifiers at neighboring point pairs but also to those at distant point pairs. Although, the dependency makes the problem non-Markovian (i.e., higher-order Markovian), it can still be solved efficiently by using a graph cut algorithm with polynomial-order computations. The experimental results revealed that the proposed method can achieve better recognition accuracy while utilizing the above characteristics of the proposed method.
    PLoS ONE 01/2013; 8(10):e76980. · 3.53 Impact Factor

Full-text (2 Sources)

View
48 Downloads
Available from
May 21, 2014