| dc.description.abstract | Currently, Particle Filters (PFs) have been applied with great success to a variety of state estimation problems. In my thesis, I propose some statistical approach to increasing the efficiency of PFs by adapting the size of sample sets during the estimation process using various methods of machine learning algorithm. In this work, I implement seven methods such as: SIR (Sequential Important Resampling), Kullback-Leibler Distance (KLD) resampling, KLD resampling adjust upper bound, KLD resampling bound error–based K-Nearest Neighbors (KNN), KLD resampling bound error –based Linear Discriminant Analysis (LDA) and KLD resampling bound error-based Support Vector Machine (SVM). All of our proposals are well-done and obtain the benefit in the average number of used particles, the performance of Root Mean Square Error (RMSE), running time, etc.
The main idea in my thesis lay a foundation for later to easier to apply many Data Science algorithms in PFs method. This code is doing well in PYTHON. Therefore, I believe that my work can apply in the real test, for example it can build on chip or TTN website. | en_US |
