1
Filtering and forecasting with B-spline networks | |
Author | Wachawee Phuetphan |
Call Number | AIT Thesis no.CS-98-8 |
Subject(s) | Neural networks (Computer science) |
Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Advanced Technologies |
Publisher | Asian Institute of Technology |
Abstract | Two B-spline network models were developed in this study for filtering and forecasting of the river discharge at several stations in the Mae Klong River basin, Thailand and the Black River basin, Vietnam. The first model was created by the combination of univariate basis functions and the second model was created by taking the tensor product of univariate basis functions. The analytical method and instantaneous learning rule were used to find the weight vector which keeps the information of the network under consideration. From the application of the models to the data at the aforementioned stations, it was found that B-spline networks can provide satisfactorily accurate filtered and forecast values, especially for the Mae Klong River basin. Both models generate somewhat similar results. However, the second model sometimes needs to use instantaneous learning rules (i.e. instantaneous gradient descent and error correction rule) for calculating the weight vector because the analytical method may lead to singular matrices which obstruct the simulation process. Moreover, the second model consumes longer calculation time. So, the first model is the preferred choice due to its simplicity and its fast calculation time. |
Year | 1998 |
Type | Thesis |
School | School of Advanced Technologies (SAT) |
Department | Other Field of Studies (No Department) |
Academic Program/FoS | Computer Science (CS) |
Chairperson(s) | Huynh Ngoc Phien; |
Examination Committee(s) | Kanchana Kanchanasut ;Hoang Le Tien; |
Scholarship Donor(s) | Electricity Generating Authority of Thailand ; |
Degree | Thesis (M.Eng.) - Asian Institute of Technology, 1998 |