# Continuous-based Heuristics for Graph and Tree Isomorphisms, with Application to Computer Vision

## Abstract

We present a new (continuous) quadratic programming approach for graph- and tree-isomorphism problems which is based on an equivalent maximum clique formulation. The approach is centered around a fundamental result proved by Motzkin and Straus in the mid-1960s, and recently expanded in various ways, which allows us to formulate the maximum clique problem in terms of a standard quadratic program. The attractive feature of this formulation is that a clear one-to-one correspondence exists between the solutions of the quadratic programs and those in the original, combinatorial problems. To approximately solve the program we use the so-called “replicator” equations, a class of straightforward continuous- and discrete-time dynamical systems developed in various branches of theoretical biology. We show how, despite their inherent inability to escape from local solutions, they nevertheless provide experimental results which are competitive with those obtained using more sophisticated mean-field annealing heuristics. Application of this approach to shape matching problems arising in computer vision and pattern recognition are also presented.

## Keywords

Maximum clique quadratic programming replicator dynamics shape recognition.## Preview

Unable to display preview. Download preview PDF.

## References

- [1]S. Arora, C. Lund, R. Motwani, M. Sudan, and M. Szegedy. Proof verification and the hardness of approximation problems. In
*Proc. 33rd Ann. Symp. Found. Comput. Sci.*, pages 14–23. Pittsburgh, PA, 1992.CrossRefGoogle Scholar - [2]L. Babai, P. Erdös, and S. M. Selkow. Random graph isomorphism.
*SIAM J. Comput.*, 9 (3): 628–635, 1980.MathSciNetzbMATHCrossRefGoogle Scholar - [3]H. G. Barrow and R. M. Burstall. Subgraph isomorphism, matching relational structures, and maximal cliques.
*Inform. Process. Lett.*, 4 (4): 83–84, 1976.zbMATHCrossRefGoogle Scholar - [4]L. E. Baum and J. A. Eagon. An inequality with applications to statistical estimation for probabilistic functions of markov processes and to a model for ecology.
*Bull. Amer. Math. Soc.*, 73: 360–363, 1967.MathSciNetzbMATHCrossRefGoogle Scholar - [5]I. M. Bomze. Evolution towards the maximum clique.
*J. Global Optim.*, 10: 143164, 1997.Google Scholar - [6]I. M. Bomze, M. Budinich, P. M. Pardalos, and M. Pelillo. The maximum clique problem. In D.-Z. Du and P. M. Pardalos, editors,
*Handbook of Combinatorial Optimization*, volume 4. Kluwer Academic Publishers, Boston, MA, 1999.Google Scholar - [7]I. M. Bomze, M. Budinich, M. Pelillo, and C. Rossi. Annealed replication: A new heuristic for the maximum clique problem.
*Discr. Appl. Math.*,1999. to appear.Google Scholar - [8]I. M. Bomze, M. Pelillo, and R. Giacomini Evolutionary approach to the maximum clique problem: Empirical evidence on a larger scale. In I. M. Bomze, T. Csendes, R. Horst, and P. M. Pardalos, editors,
*Developments in Global Optimization*, pages 95–108. Kluwer Academic Publishers, Dordrecht, The Netherlands, 1997.CrossRefGoogle Scholar - [9]R. B. Boppana, J. Hastad, and S. Zachos. Does co-NP have short interactive proofs?
*Inform. Process. Lett.*, 25: 127–132, 1987.MathSciNetzbMATHCrossRefGoogle Scholar - [10]R. Brockett and P. Maragos. Evolution equations for continuous-scale morphology. In
*Proceedings of the IEEE Conference on Acoustics, Speech and Signal Processing*, San Francisco, CA, March 1992.Google Scholar - [11]R. Durbin and D. Willshaw. An analog approach to the travelling salesman problem using an elastic net method.
*Nature*, 326: 689–691, 1987.CrossRefGoogle Scholar - [12]Y. Fu and P. W. Anderson. Application of statistical mechanics to NP-complete problems in combinatorial optimization.
*J. Phys. A*, 19: 1605–1620, 1986.MathSciNetzbMATHCrossRefGoogle Scholar - [13]M. R. Garey and D. S. Johnson.
*Computers and Intractability: A Guide to the Theory of NP-Completeness. W*. H. Freeman, San Francisco, CA, 1979.Google Scholar - [14]L. E. Gibbons, D. W. Hearn, and P. M. Pardalos. A continuous based heuristic for the maximum clique problem. In D. S. Johnson and M. Trick, editors,
*Cliques, Coloring, and Satisfiability—Second DIMACS Implementation Challenge*, pages 103–124. American Mathematical Society, 1996.Google Scholar - [15]L. E. Gibbons, D. W. Hearn, P. M. Pardalos, and M. V. Ramana. Continuous characterizations of the maximum clique problem.
*Math. Oper. Res.*, 22: 754–768, 1997.MathSciNetzbMATHCrossRefGoogle Scholar - [16]S. Gold and A. Rangarajan. A graduated assignment algorithm for graph matching.
*IEEE Trans. Pattern Anal. Machine Intell.*, 18 (4): 377–388, 1996.CrossRefGoogle Scholar - [17]M. Grötschel, L. Lovâsz, and A. Schrijver.
*Geometric Algorithms and Combinatorial Optimization*. Springer-Verlag, Berlin, 1988.zbMATHCrossRefGoogle Scholar - [18]F. Harary.
*Graph Theory*. Addison-Wesley, Reading, MA, 1969.Google Scholar - [19]J. Hastad. Clique is hard to approximate within
*n*1—E. In*Proc. 37th Ann. Symp. Found. Comput. Sci.*, pages 627–636, 1996.Google Scholar - [20]J. Hofbauer. Imitation dynamics for games. Collegium Budapest, preprint, 1995.Google Scholar
- [21]J. Hofbauer and K. Sigmund.
*The Theory of Evolution and Dynamical Systems*. Cambridge University Press, Cambridge, UK, 1988.zbMATHGoogle Scholar - [22]J. J. Hopfield and D. W. Tank. Neural computation of decisions in optimization problems.
*Biol. Cybern.*, 52: 141–152, 1985.MathSciNetzbMATHGoogle Scholar - [23]A. Jagota. Approximating maximum clique with a Hopfield network.
*IEEE Trans. Neural Networks*, 6: 724–735, 1995.CrossRefGoogle Scholar - [24]D. S. Johnson. The NP-completeness column: An ongoing guide.
*J. Algorithms*, 9: 426–444, 1988.MathSciNetzbMATHCrossRefGoogle Scholar - [25]B. B. Kimia, A. Tannenbaum, and S. W. Zucker. Shape, shocks, and deformations I: The components of two-dimensional shape and the reaction-diffusion space.
*Int. J. Comp. Vision*, 15: 189–224, 1995.CrossRefGoogle Scholar - [26]J. J. Kosowsky and A. L. Yuille. The invisible hand algorithm: Solving the assignment problem with statistical physics.
*Neural Networks*, 7: 477–490, 1994.zbMATHCrossRefGoogle Scholar - [27]D. Kozen. A clique problem equivalent to graph isomorphism.
*SIGACT News*, pages 50–52, Summer 1978.Google Scholar - [28]P. D. Lax. Shock waves and entropy. In E. H. Zarantonello, editor,
*Contributions to Nonlinear Functional Analysis*, pages 603–634, New York, 1971. Acad. Press.Google Scholar - [29]J. T. Li, K. Zhang, K. Jeong, and D. Shasha. A system for approximate tree matching.
*IEEE Trans. Knowledge Data Eng.*, 6: 559–571, 1994.CrossRefGoogle Scholar - [30]V. Losert and E. Akin. Dynamics of games and genes: Discrete versus continuous time.
*J. Math. Biol.*, 17: 241–251, 1983.MathSciNetzbMATHCrossRefGoogle Scholar - [31]S. Y. Lu. A tree-matching algorithm based on node splitting and merging.
*IEEE Trans. Pattern Anal. Machine Intell.*, 6: 249–256, 1984.zbMATHCrossRefGoogle Scholar - [32]D. Marr and K. H. Nishihara. Representation and recognition of the spatial organization of three-dimensional shapes.
*Proc. R. Soc. Lond. B*, 200: 269–294, 1978.CrossRefGoogle Scholar - [33]D. W. Matula. An algorithm for subtree identification.
*Siam Rev*., 10: 273–274, 1968.Google Scholar - [34]T. S. Motzkin and E. G. Straus. Maxima for graphs and a new proof of a theorem of Turd’’’.
*Canad. J. Math.*, 17: 533–540, 1965.MathSciNetzbMATHCrossRefGoogle Scholar - [35]M. Neff, R. Byrd, and O. Rizk. Creating and querying hierarchical lexical databases. In
*Proc. 2nd Conf. Appl. Natural Language Process.*, pages 84–93, 1988.CrossRefGoogle Scholar - [36]M. Ohlsson, C. Peterson, and B. Söderberg. Neural networks for optimization problems with inequality constraints: The knapsack problem.
*Neural Computation*, 5: 331–339, 1993.CrossRefGoogle Scholar - [37]E. M. Palmer.
*Graphical Evolution: An Introduction to the Theory of Random Graphs*. John Wiley & Sons, New York, 1985.zbMATHGoogle Scholar - [38]P. M. Pardalos. Continuous approaches to discrete optimization problems. In G. D. Pillo and F. Giannessi, editors,
*Nonlinear Optimization and Applications*, pages 313–328. Plenum Press, 1996.Google Scholar - [39]P. M. Pardalos and A. T. Phillips. A global optimization approach for solving the maximum clique problem.
*Int. J. Comput. Math.*, 33: 209–216, 1990.zbMATHCrossRefGoogle Scholar - [40]M. Pelillo. Relaxation labeling networks for the maximum clique problem.
*J. Artif. Neural Networks*, 2: 313–328, 1995.Google Scholar - [41]M. Pelillo. Replicator equations, maximal cliques, and graph isomorphism.
*Neural Computation*, 11 (8): 2023–2045, 1999.CrossRefGoogle Scholar - [42]M. Pelillo and A. Jagota. Feasible and infeasible maxima in a quadratic program for maximum clique.
*J. Artif. Neural Networks*, 2: 411–420, 1995.Google Scholar - [43]M. Pelillo, K. Siddiqi, and S. W. Zucker. Attributed tree matching and maximum weight cliques. In
*Proc. ICIAP’99–10th Int. Conf. on Image Analysis and Processing*. IEEE Computer Society Press, 1999.Google Scholar - [44]A. Rangarajan, S. Gold, and E. Mjolsness. A novel optimizing network architecture with applications.
*Neural Computation*, 8: 1041–1060, 1996.CrossRefGoogle Scholar - [45]A. Rangarajan and E. Mjolsness. A lagrangian relaxation network for graph matching.
*IEEE Trans. Neural Networks*, 7 (6): 1365–1381, 1996.CrossRefGoogle Scholar - [46]S. W. Reyner. An analysis of a good algorithm for the subtree problem.
*SIAM J. Comput.*, 6: 730–732, 1977.MathSciNetzbMATHCrossRefGoogle Scholar - [47]H. Rom and G. Medioni. Hierarchical decomposition and axial shape description.
*IEEE Trans. Pattern Anal. Machine Intell.*, 15 (10): 973–981, 1993.CrossRefGoogle Scholar - [48]H. Samet.
*Design and Analysis of Spatial Data Structures*. Addison-Wesley, Reading, MA, 1990.Google Scholar - [49]U. Schöning. Graph isomorphism is in the low hierarchy.
*J. Comput. Syst. Sci.*, 37: 312–323, 1988.zbMATHCrossRefGoogle Scholar - [50]B. A. Shapiro and K. Zhang. Comparing multiple RNA secondary structures using tree comparisons.
*Comput. Appl. Biosci.*, 6: 309–318, 1990.Google Scholar - [51]D. Shasha, J. T. L. Wang, K. Zhang, and F. Y. Shih. Exact and approximate algorithms for unordered tree matching.
*IEEE Trans. Syst. Man Cybern.*, 24: 668678, 1994.Google Scholar - [52]K. Siddiqi, A. Shokoufandeh, S. J. Dickinson, and S. W. Zucker. Shock graphs and shape matching.
*Int. J. Comp. Vision*, to appear, 1999.Google Scholar - [53]P. D. Simié. Constrained nets for graph matching and other quadratic assignment problems.
*Neural Computation*, 3: 268–281, 1991.CrossRefGoogle Scholar - [54]J. W. Weibull.
*Evolutionary Game Theory*. MIT Press, Cambridge, MA, 1995.zbMATHGoogle Scholar - [55]H. S. Wilf. Spectral bounds for the clique and independence numbers of graphs.
*J. Combin. Theory*,*Ser. B*, 40: 113–117, 1986.MathSciNetzbMATHGoogle Scholar - [56]S. Zhu and A. L. Yuille. FORMS: A flexible object recognition and modeling system.
*Int. J. Comp. Vision*, 20 (3): 187–212, 1996.CrossRefGoogle Scholar