A discriminative learning framework with pairwise constraints for video object classification

To deal with the problem of insufficient labeled data in video object classification, one solution is to utilize additional pairwise constraints that indicate the relationship between two examples, i.e., whether these examples belong to the same class or not. In this paper, we propose a discriminative learning approach which can incorporate pairwise constraints into a conventional margin-based learning framework. Different from previous work that usually attempts to learn better distance metrics or estimate the underlying data distribution, the proposed approach can directly model the decision boundary and, thus, require fewer model assumptions. Moreover, the proposed approach can handle both labeled data and pairwise constraints in a unified framework. In this work, we investigate two families of pairwise loss functions, namely, convex and nonconvex pairwise loss functions, and then derive three pairwise learning algorithms by plugging in the hinge loss and the logistic loss functions. The proposed learning algorithms were evaluated using a people identification task on two surveillance video data sets. The experiments demonstrated that the proposed pairwise learning algorithms considerably outperform the baseline classifiers using only labeled data and two other pairwise learning algorithms with the same amount of pairwise constraints.     

Learning vector quantization with training data selection

In this paper, we propose a method that selects a subset of the training data points to update LVQ prototypes. The main goal is to conduct the prototypes to converge at a more convenient location, diminishing misclassification errors. The method selects an update set composed by a subset of points considered to be at the risk of being captured by another class prototype. We associate the proposed methodology to a weighted norm, instead of the Euclidean, in order to establish different levels of relevance for the input attributes. The technique was implemented on a controlled experiment and on Web available data sets.     

From the editors

Proceedings of the 7th International Conference on Pattern Recognition and Image Analysis: New Information Technologies (PRIA-7-2004)18–23 October, 2004, St. Petersburg, Russian Federation Revised papers

From the editors     

Exploiting symmetry in the synthesis of supervisors for discreteevent systems

We investigate a method to exploit the symmetry inherent to many discrete-event systems in order to reduce the computational complexity of the supervisory control problem (SCP). We characterize symmetry using notions of group theory and derive conditions under which there exists a quotient (reduced) automaton representation for a language with, in general, a much smaller state space than the original minimal (in number of states) automaton representation for the language. We then propose an algorithm to synthesize a solution to the SCP which is similar to the classical one, but performed on reduced automata. Special attention is given to the particular case of systems whose models contain similar components. The approach is illustrated by an example of control of a small production line     

Acceleration and filtering in the generalized Landweber iteration using a variable shaping matrix

The generalized Landweber iteration with a variable shaping matrix is used to solve the large linear system of equations arising in the image reconstruction problem of emission tomography. The method is based on the property that once a spatial frequency image component is almost recovered within in in the generalized Landweber iteration, this component will still stay within in during subsequent iterations with a different shaping matrix, as long as this shaping matrix satisfies the convergence criterion for the component. Two different shaping matrices are used: the first recovers low-frequency image components; and the second may be used either to accelerate the reconstruction of high-frequency image components, or to attenuate these components to filter the image. The variable shaping matrix gives results similar to truncated inverse filtering, but requires much less computation and memory, since it does not rely on the singular value decomposition.     

Use of TVD-schemes to simulate gasdynamic flows with spherical and cylindrical symmetry

A generalization of the Harten TVD-scheme in the form of a predictor-corrector-type scheme of second-order accuracy in time and space is proposed for the flows with spherical and cylindrical symmetry. The scheme does not require introducing components with artificial viscosity and describes discontinuities in the solution without the appearance of nonphysical oscillations for a Courant number less than 0.8.Academic Scientific Complex A. V. Luikov Heat and Mass Transfer Institute of the Academy of Sciences of Belarus, Minsk. Translated from Inzhenerno-Fizicheskii Zhurnal, Vol. 65, No. 3, pp. 296–301, September, 1993.     

Phylogeny of trichomonads based on partial sequences of large subunit rRNA and on cladistic analysis of morphological data

Several domains of large subunit rRNA from nine trichomonad species have been sequenced. Molecular phylogenies obtained with parsimony and distance methods demonstrate the trichomonads are a monophyletic group which branches very early in the eukaryotic tree. The topology of the trees is in general agreement with traditional views on evolutionary and systematic relationships of trichomonads. A clear dichotomy is noted between the subfamily Trichomonadinae and the subfamily Tritrichomonadinae. In the latter subfamily, a second division separates the "Tritrichomonas muris-type" species from the "Tritrichomonas augusta-type" ones. Previous evolutionary schemes in which the Monocercomonadidae were regarded as the most "primitive" and the Trichomonadidae as more "evolved" are not in agreement with our molecular data. The emergence of Monocercomonas and Hypotrichomonas at the base of the Tritrichomonas lineage suggests a secondary loss of some cytoskeletal structures, the costa and undulating membrane in these genera. This is corroborated by the early branching position of Trichomitus, which possesses a costa and an undulating membrane and has usually been placed among the Trichomonadidae on the basis of cytological characters. A cladistic analysis was applied to the available morphological characters in order to produce a hierarchical grouping of the taxa reflecting their morphological diversity.