1
Constrained topological mapping for regression analysis | |
Author | Myint Myint Moe |
Call Number | AIT Thesis no. CS-92-14 |
Subject(s) | Regression analysis |
Note | A Thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Engineering and Technology |
Publisher | Asian Institute of Technology |
Abstract | The Optimal knot placement is very important in the role of piecewise polynomial function approximation. The classical approach to the optimal knot location is a computionally hard problem of combinatorial complexity even for single-variable functions. In this thesis the idea of using Kohonen' s self-organizing maps is applied to the problem of regression analysis with preserving the topological ordering of projections of units onto (N-1) dimensional subspace of independent coordinates. The output units have correct topological order in the mapping of giving the number of data points in N-dimensional input space. The value of the function among the units of mature map is found, in principle, by piecewise polynomial approximation. The developed algorithm produced optimal placement of knots and the response values of immediate data points in very reasonable processing time. Simulation examples illustrate the performance of the proposed algorithm for approximating one and two variable functions in different parametric and problem conditions. |
Year | 1992 |
Type | Thesis |
School | School of Engineering and Technology (SET) |
Department | Department of Information and Communications Technologies (DICT) |
Academic Program/FoS | Computer Science (CS) |
Chairperson(s) | Sadananda, Ramakoti |
Examination Committee(s) | Huynh, Ngoc Phien ;Charoon Chirapaisamkul |
Scholarship Donor(s) | Royal Government of Thailand ; |
Degree | Thesis (M.Eng.) - Asian Institute of Technology, 1992 |