A further study on backpropagation and B-spline neural networks in filtering and forecasting problems | |
| Author | Nguyen Duc Anh Kha |
| Call Number | AIT Thesis no.IM-01-05 |
| Subject(s) | Pattern recognition systems Optical character recognition devices Telephone, Wireless |
| 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 |
| Series Statement | Thesis ; no. IM-01-05 |
| Abstract | Artificial Neural Networks are applied widely in many fields in recent years in filtering and forecasting non-linear time series data. In this study, two new neural network models are considered: the Backpropagation and the B-spline models. Two computer software packages based on these two models were implemented to assess the capabilities of these models in filtering and forecasting of time series data. Moreover, for the Backpropagation model, two new algorithms were employed to improve the convergence speed of the model and the Bayesian Information Criterion (BIC) was employed to find the optimal network structure. Two data sets, one with discharge and water level of the Thu Bon River, Vietnam, and the other with discharge of the Mae Klong River, Thailand, were used to evaluate the efficiency of the software packages. The results obtained from these models show that both Backpropagation and B-spline models can be applied satisfactorily in filtering and forecasting problems. Moreover, the Backpropagation model can achieve a higher accuracy than the B-spline model, although it requires a longer computation time. |
| Year | 2001 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis ; no. IM-01-05 |
| Type | Thesis |
| School | School of Advanced Technologies (SAT) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Information Management (IM) |
| Chairperson(s) | Huynh Ngoc Phien; |
| Examination Committee(s) | Hoang Le Tien;Sujarae, Aekavute |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology, 2001 |