Vai al contenuto| Home page|

   Ti trovi in: HOME »Programmi, progetti e risultati »I progetti »PRIN - Programmi di ricerca di Rilevante Interesse Nazionale»Programma di ricerca
INIZIO_TESTO_DA_INDICIZZARE

RESEARCH PROGRAM

italiano - inglese
Research Units
Similar research programs:
Scientific and education field classification
International Patent Classification
  • PHYSICS
    • COMPUTING; CALCULATING; COUNTING (score computers for games A63; combinations of writing applicances with computing devices B43K29/08)
      • COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS [N0004]
      • IMAGE DATA PROCESSING OR GENERATION, IN GENERAL (specially adapted for particular applications, see the relevant subclasses, e.g. G06K, G09G, H04N) [N9408]
    • MUSICAL INSTRUMENTS; ACOUSTICS
      • SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION (measurement of sound waves in general G01H; frequency spectrum analysis of electric signals in general G01R23/16; sound input/output for computers G06F3/16; computing specially adapted for specific functions G06F17/00, G06G7/00; image data processing G06T; teaching or communicating with the blind, deaf or mute G09B; information storage, e.g. sound storage, G11; electronic circuits for sound generation H03B; electronic filters H03H; coding, decoding or code conversion, in general H03M; transmission of information, e.g. speech, H04B; telephonic communication H04M; microphone arrangements, hearing aids, public address systems H04R) [C9607]
Geographical classification
Bibliografia
[AKA91] D. Aha, D. Kibler, M. Albert. Instance-based learning algorithms. Mach. Learning 6:37-66 (1991).

[B97] H. Bunke. On a relation between graph edit distance and maximum common subgraph. Pattern Recognition Lett. 18:689-694 (1997).

[BB05] M. Basu, H. Bunke (Eds.) IEEE Trans. PAMI. Special issue on "Syntactic and structural pattern recognition", 27 (2005).

[BB76] H. G. Barrow, R. M. Burstall. Subgraph isomorphism, matching relational structures and maximal cliques. Inform. Process. Lett. 4:83-84 (1976).

[BC04] H. Bunke, T. Caelli (Eds.) Int. J. Pattern Recognition Artif. Intell. Special issue on "Graph matching in pattern recognition and machine vision" 18 (2004).

[BK88] K. L. Boyer, A. C. Kak. Structural stereopsis for 3-D vision. IEEE Trans. PAMI 10:144-166 (1988).

[BMPT06] M. Bicego, V. Murino, M. Pelillo, A. Torsello (Eds.). Pattern Recognition special issue on "Similarity based pattern recognition" (in press).

[BOP97] M. Brand, N. Oliver, S. Pentland. "Coupled hidden Markov models for complex action recognition". In Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 994-99 (1997).

[BS98] H. Bunke, K. Shearer. A graph distance metric based on the maximal common subgraph. Pattern Recognition Lett. 19:255-259 (1998).

[CH67] Cover, T.M., Hart, P.E. Nearest neighbor pattern classification. IEEE Trans. Inf. Theory 13:21-27 (1967).

[CKP95] W. J. Christmas, J. Kittler, M. Petrou. Structural matching in computer vision using probabilistic relaxation. IEEE Trans. PAMI 17:749-764 (1995).

[DHS01] R. O. Duda, P. E. Hart, D. G. Stork. Pattern Classification. J. Wiley & Sons, New York, 2001.

[DPZ01] S. Dickinson, M. Pelillo, R. Zabih (Eds.). IEEE Trans. PAMI. Special issue on "Graph algorithms in computer vision" 23(10) (2001).

[E99] S. Edelman. Representation and Recognition in Vision. MIT Press, Cambridge, 1999.

[E86] M. A. Eshera, K. S. Fu. An image understanding system using attributed symbolic representation and inexact graph-matching. IEEE Trans. PAMI 8:604-618 (1986).

[F90] K. Fukunaga. Introduction to Statistical Pattern Recognition. Academic Press, New York, 1990.

[FB03] B. Fischer, J. M. Buhmann. Path-based clustering for grouping smooth curves and texture segmentation. IEEE Trans. PAMI 25:513-518 (2003).

[FST98] S. Fine, Y. Singer, N. Tishby. The hierarchical hidden Markov model: Analysis and applications. Machine Learning, 32:41-62 (1998).

[G99] R. L. Goldstone. Similarity. In R. A. Wilson, F. C. Keil (Eds.) MIT Encyclopedia of the Cognitive Sciences. MIT Press, Cambridge, MA, pp. 763-765 (1999).

[G+99] T. Graepel, R. Herbrich, P. Bollmann-Sdorra, K. Obermayer. Classification on pairwise proximitydata. In D. Cohn, M. Kearns, S. Solla (Eds.), Advances in NIPS 11:438-444 (1999).

[GR05] G. Giacinto, F. Roli. Instance-Based Relevance Feedback for Image Retrieval. In L.K. Saul, Y. Weiss, and Leon Bottou (Eds.) Advances in NIPS 17:489-496 (2005).

[HB97] T. Hofmann, J. Buhmann. Pairwise data clustering by deterministic annealing. IEEE Trans. PAMI 19:1-14, (1997).

[HH93] L. Herault, R. Horaud. Figure-ground discrimination: A combinatorial optimization approach. IEEE Trans. PAMI 15:899-914 (1993).

[HSSS] P.J. Green, N.L. Hjort, S. Richardson (Eds.). Highly Structured Stochastic Systems, vol. 27 of Oxford Statistical Science Series, OUP, Oxford, UK, 2003.

[JD88] A. K. Jain, R. C. Dubes. Algorithms for Clustering Data. Prentice Hall, Englewood Cliffs, NJ, 1988.

[JW00] D.W. Jacobs, D. Weinshall, Classification with nonmetric distances: image retrieval and class representation, IEEE Trans. PAMI 22:583-600 (2000).

[JZ97] A. K. Jain, D. Zongker, Representation and recognition of handwritten digits using deformable templates, IEEE Trans. PAMI 19:1386-1391 (1997)

[KSS03] J. Keuchel, C. Schnorr, C. Schellewald, D. Cremers. Binary partitioning, perceptual grouping, and restoration with semidefinite programming. IEEE Trans. PAMI 25:1364-1379 (2003).

[L96] S.L. Lauritzen. Graphical Models, Oxford Science Publications, 1996.

[P97] J. Pearl. Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference, Morgan Kaufmann, 1997.

[PD01] E. Pekalska, R.P.W. Duin, Automatic pattern recognition by similarity representations, Electron. Lett. 37:159-160 (2001).

[PD02] E. Pekalska, R.P.W. Duin, Dissimilarity representations allow for building good classifiers, Pattern Recogn. Lett. 23:943-956 (2002).

[PPD02] E. Pekalska, P. Paclik, R.P.W. Duin, A generalized kernel approach to dissimilarity-based classification, JMLR 2:175-211 (2002).

[PF98] P. Perona, W. Freeman. A factorization approach to grouping. In Proc. ECCV'98, pp. 655-670. Springer, Berlin (1998).

[R89] L.R. Rabiner. A tutorial on hidden Markov models and selected applications in speech recognition. Proc. IEEE, 77:257-286 (1989).

[RY81] V. V. Raghavan and C. T. Yu. A comparison of the stability characteristics of some graph theoretic clustering methods. IEEE Trans. PAMI 3:393-402 (1981).

[SB98] S. Sarkar and K. L. Boyer. Quantitative measures of change based on feature organization: Eigenvalues and eigenvectors. CVIU 71:110-136 (1998).

[SH82] L. G. Shapiro and R. M. Haralick. Relational models for scene analysis. IEEE Trans. PAMI, 4:595-602 (1982).

[SM00] J. Shi and J. Malik. Normalized cuts and image segmentation. IEEE Trans. PAMI 22:888-905 (2000).

[T+06a] D. Tao, X. Tang, X. Li, and X. Wu. Asymmetric Bagging and Random Subspace for Support Vector Machines-based Relevance Feedback in Image Retrieval. IEEE Trans. PAMI, (to appear 2006).

[T+06b] D. Tao, X. Tang, X. Li, Y. Rui. Direct Kernel Biased Discriminant Analysis: A New Content-based Image Retrieval Relevance Feedback Algorithm. IEEE Trans. Multimedia, (to appear 2006).

[V98] V. Vapnik, Statistical Learning Theory, Wiley, NewYork, 1998.

[W+92] C. Wharton et al. The story with reminding: memory retrieval is influenced by analogical similarity. In Proc. 14th Annual Conf. of the Cognitive Science Society, Bloomington, IN, pp. 588-593 (1992).

[WH97] R. Wilson, E. R. Hancock. Structural matching by discrete relaxation. IEEE Trans. PAMI 19:634-648 (1997).

[WL93] Z .Wu and R. Leahy. An optimal graph theoretic approach to data clustering: Theory and its application to image segmentation. IEEE Trans. PAMI 15:1101-1113 (1993).

[WSKR01] W. D. Wallis, P. Shoubridge, M. Kraetz, and D. Ray. Graph distances using graph union. Pattern Recognition Lett., 22:701-704 (2001).

[WY85] A. K. C. Wong and M. You. Entropy and distance of random graphs with application to structural pattern recognition. IEEE Trans. PAMI 7:599-609 (1985).

[Z71] C. T. Zahn. Graph-theoretic methods for detecting and describing gestalt clusters. IEEE Trans. Comput. 20:68-86 (1971).

[ZH01] Zhou XS and Huang TS. Small sample learning during multimedia retrieval using BiasMap. IEEE Conf. Computer Vision and Pattern Recognition, 2001.

[ZH03] X. S. Zhou, T.S. Huang. Relevance Feedback in Image Retrieval: A Comprehensive Review. Multimedia Systems 8:536-544 (2003).

[ZSS92] K. Zhang, R. Statman, and D. Shasha. On the editing distance between unordered labeled trees. Inform. Process. Lett. 42:133-139 (1992).
Keywords
PATTERN RECOGNITION, COMPUTER VISION, SIMILARITY, GRAPHS AND RELATIONAL STRUCTURES, PROBABILISTIC GRAPHICAL MODELS, GENERATIVE AND DISCRIMINATIVE MODELS, FACE DETECTION AND RECOGNITION, CONTENT-BASED IMAGE RETRIEVAL

Similarity-based Methods for Computer Vision and Pattern Recognition: Theory, Algorithms, Applications

Università "Ca' Foscari" di Venezia
Abstract
Traditional pattern recognition techniques are centered around the notion of "feature." According to this view, the objects to be classified are represented in terms of properties that are intrinsic to the object itself. Hence, a typical pattern recognition system makes its decisions by simply looking at one or more feature vectors fed as input. The strength of this approach is that it can leverage a wide range of mathematical tools ranging from statistics, to geometry, to optimization. However, in many real-world applications the objects are not naturally representable in terms of a vector of features. For example, graph (or structural) representations lack a canonical order or correspondence between nodes. On the other hand, quite often it is possible to obtain a measure of the (dis)similarity of the objects to be classified. It is therefore tempting to design a pattern recognizer which, unlike traditional systems, accepts as input a matrix containing the similarities between objects and produces class labels as output. This allows one to develop algorithms that are independent from the actual data representation, allowing the use of non-metric similarities. Further, this makes the approaches applicable to problems that do not have a natural embedding to a uniform feature space, such as the clustering of structural or graph-based representations or the analysis of sequences. These representations are well suited to both supervised and unsupervised >>>

Principal Investigator
Marcello Pelillo Università "Ca' Foscari" di VENEZIA
Research Objectives
The challenge of automatic pattern recognition is to develop computational methods which learn to distinguish among a number of classes represented by examples. Traditional pattern recognition techniques are centered around the notion of “feature” [DHS01]. According to this view, the objects to be classified are represented in terms of properties that are intrinsic to the object itself. Hence, a typical pattern recognition system makes its decisions by simply looking at one or more feature vectors fed as input. The strength of this approach is that it can leverage a wide range of mathematical tools ranging from statistics, to geometry, to optimization.

However, in many real-world applications a feasible feature-based description of objects might be difficult to obtain or inefficient for learning purposes. This is typically the case when experts cannot define features in a straightforward way, when data are high dimensional, when features consist of both continuous and categorical variables, or when the objects to be classified are represented in terms of graphs or structural representations. In these cases, it is often possible to obtain a measure of the (dis)similarity of the objects to be classified, and in some applications, e.g., shape recognition [E99], the use of dissimilarities (rather than features) makes the problem more viable. It is therefore tempting to design a pattern recognizer which, unlike traditional systems, accepts as input a matrix >>>

Timescale
24 months
National and international background
Similarity-based classification has been a recurring yet understudied theme in the pattern recognition literature for over 40 years. A straightforward approach to dissimilarity representations to supervised classification leads to the nearest neighbor rule [CH67,F90] or more generally to instance-based learning [AKA91]. Such classifiers make use of distance information in a rank-based way. The NN rule, in its simplest form (1-NN), assigns a new object to the class with the nearest representation element. The k-NN decision rule is based on majority voting: an unknown object becomes a member of the class most frequently occurring among the k-NN.

The NN rule has found application is several domains. For example, Jain and Zongker [JZ97] apply it to hand-written digit recognition. Here a dissimilarity measure is obtained from deformable templates and a 1-NN algorithm is used to perform the final classification of the patterns.

Although the decision rule of NN methods is based on local neighborhoods, it is still computationally expensive, since dissimilarities to all training examples have to be found. To overcome such limitations, Pekalska and Duin [PD02] recently proposed to train the classifier on the dissimilarities to a set of prototypes, called the representation set. Graepel et al. [G+99] investigate the problem of learning a classifier based on data represented in terms of their pairwise proximities, using an approach based on Vapnik's >>>