On interval weighted three-layer neural networks

When solving application problems, the data sets used to train a neural network may not be one hundred percent precise but are within a certain range. By representing data sets with intervals, one has interval neural networks. By analyzing the mathematical model, the authors categorize general three-layer neural network training problems into two types. One of them can be solved by finding numerical solutions of nonlinear systems of equations. The other can be transformed into nonlinear optimization problems. Reliable interval algorithms such as interval Newton/generalized bisection method and interval branch-and-bound algorithm are applied to obtain optimal weights for interval neural networks. Applicable state-of-art interval software packages are also reviewed.

[1]  Chenyi Hu,et al.  On Parallelization of Interval Newton Method on Distributed-Memory Multiprocessors , 1993, SIAM Conference on Parallel Processing for Scientific Computing.

[2]  Chenyi Hu,et al.  Algorithm 737: INTLIB—a portable Fortran 77 interval standard-function library , 1994, TOMS.

[3]  Qing Yang,et al.  Optimizing INTBIS on the CRAY Y-MP , 1995, Reliab. Comput..

[4]  Hu Chengyi,et al.  Parallel solutions for large-scale general sparse nonlinear systems of equations , 1996 .

[5]  E. Hansen,et al.  Bounding solutions of systems of equations using interval analysis , 1981 .

[6]  Eldon Hansen,et al.  Global optimization using interval analysis , 1992, Pure and applied mathematics.

[7]  Ramon E. Moore.  Methods and applications of interval analysis , 1979, SIAM studies in applied mathematics.

[8]  Ronald R. Willis,et al.  New Computer Methods for Global Optimization , 1990 .

[9]  R. B. Kearfott,et al.  A Review of Preconditioners for the Interval Gauss–Seidel Method , 2000 .

[10]  R. B. Kearfott,et al.  Applications of interval computations , 1996 .

[11]  Ramon E. Moore.  Interval arithmetic and automatic error analysis in digital computing , 1963 .

[12]  S. Thomas Alexander,et al.  Adaptive Signal Processing , 1986, Texts and Monographs in Computer Science.

[13]  R. B. Kearfott.  Rigorous Global Search: Continuous Problems , 1996 .

[14]  E. Hansen.  On solving systems of equations using interval arithmetic , 1968 .

[15]  R. B. Kearfott,et al.  Optimal preconditioners for the interval newton method , 1990 .

[16]  胡承毅.  Parallel Solutions for Large-Scale General Sparse Nonlinear Systems of Equations , 1996 .

[17]  Qing Yang,et al.  A general iterative sparse linear solver and its parallelization for interval Newton methods , 1995, Reliab. Comput..

[18]  R. Baker Kearfott,et al.  A Fortran 90 environment for research and prototyping of enclosure algorithms for nonlinear equations and global optimization , 1995, TOMS.

[20]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[21]  G. Alefeld,et al.  Introduction to Interval Computation , 1983 .

[22]  Vladik Kreinovich,et al.  3-layer Neural Networks Are Universal Approximations for Functions and for Control Strategies , 1993, Neural Parallel Sci. Comput..

[23]  R. Baker Kearfott,et al.  Preconditioners for the interval Gauss-Seidel method , 1990 .

[24]  R. B. Kearfott,et al.  Abstract generalized bisection and a cost bound , 1987 .

[25]  R. Baker Kearfott,et al.  A Review of Techniques in the Verified Solution of Constrained Global Optimization Problems , 1996 .

[26]  Chenyi Hu.  A Parallel Algorithm for Interval Polynomial Interpolation , 1997, PPSC.