Covariate Shift Adaptation by Importance Weighted Cross Validation
The Journal of Machine Learning Research
Improved computation for Levenberg-Marquardt training
IEEE Transactions on Neural Networks
Analyzing localization errors in one-dimensional sensor networks
Signal Processing
Hi-index | 0.08 |
The 3D measurements of the human body surface or anatomical areas have gained importance in many medical applications. Three dimensional laser scanning systems can provide these measurements; however usually these scanners have non-linear variations in their measurement, and typically these variations depend on the position of the scanner with respect to the person. In this paper, the Levenberg-Marquardt method is used as a digital rectifier to adjust this non-linear variation and increases the measurement accuracy of our 3D Rotational Body Scanner. A comparative analysis with other methods such as Polak-Ribire and quasi-Newton method, and the overall system functioning is presented. Finally, computational experiments are conducted to verify the performance of the proposed system and its method uncertainty.