SVM learning with the Schur-Hadamard inner product for graphs
نویسندگان
چکیده
We apply support vector learning to attributed graphs where the kernel matrices are based on approximations of the Schur-Hadamard inner product. The evaluation of the Schur-Hadamard inner product for a pair of graphs requires the determination of an optimal match between their nodes and edges. It is therefore efficiently approximated by means of recurrent neural networks. The optimal mapping involved allows a direct understanding of the similarity or dissimilarity of the two graphs considered. We present and discuss experimental results of different classifiers constructed by a SVM operating on positive semi-definite (psd) and non-psd kernel matrices.
منابع مشابه
On a Schur complement inequality for the Hadamard product of certain totally nonnegative matrices
Under the entrywise dominance partial ordering, T.L. Markham and R.L. Smith obtained a Schur complement inequality for the Hadamard product of two tridiagonal totally nonnegative matrices. Applying the properties of the Hadamard core of totally nonnegative matrices, the Schur complement inequalities for the Hadamard product of totally nonnegative matrices is obtained, which extends those of T.L...
متن کاملEla on a Schur Complement Inequality for the Hadamard Product of Certain Totally Nonnegative Matrices
Under the entrywise dominance partial ordering, T.L. Markham and R.L. Smith obtained a Schur complement inequality for the Hadamard product of two tridiagonal totally nonnegative matrices. Applying the properties of the Hadamard core of totally nonnegative matrices, the Schur complement inequalities for the Hadamard product of totally nonnegative matrices is obtained, which extends those of T.L...
متن کاملAlternating Schur Series and Necessary Conditions for the Existence of Strongly Regular Graphs
Considering the Euclidean Jordan algebra of the real symmetric matrices endowed with the Jordan product and the inner product given by the usual trace of matrices, we construct an alternating Schur series with an element of the Jordan frame associated to the adjacency matrix of a strongly regular graph. From this series we establish necessary conditions for the existence of strongly regular gra...
متن کاملOn the Hadamard product of inverse M-matrices
We investigate the Hadamard product of inverse M-matrices and present two classes of inverse M-matrices that are closed under the Hadamard multiplication. In the end, we give some inequalities on the Fan product of M-matrices and Schur complements. © 2000 Elsevier Science Inc. All rights reserved. AMS classification: 15A09; 15A42
متن کاملGeneralized Esprit Algorithm Based on Schur-hadamard Product for Doa Estimation of General-rank Signals
A generalized directions-of-arrival (DOAs) estimation algorithm is developed for the rank-one (point source) and higher-rank (distributed source) signal models. For higher-rank signals, the integral steering vector is deduced to be a Schur-Hadamard product comprising the steering vector of rank-one source and a real matrix according to the symmetry assumption of angular signal intensity. And th...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید
ثبت ناماگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید
ورودعنوان ژورنال:
- Neurocomputing
دوره 64 شماره
صفحات -
تاریخ انتشار 2005