Robust classification of objects using a sparse detector sensor
Abstract
This paper emphasizes techniques for broad-scale classification of objects sensed by a prototype unattended ground sparse detector sensor. A wide range of machine learning techniques were applied to create models of three broad classes of objects, such as humans, humans wearing large backpacks, and non-humans using data obtained from the sparse detector sensor in a laboratory environment. Fusion of models was performed based on a measure of diversity among classifiers to improve the robustness and also the accuracy of the models. Empirical analysis on 230 sample datasets shows up to a 91.74% accuracy (10-fold cross validation) in classifying three broad classes of objects of interest and shows very promising scores on other various performance indices.
Publication Title
Proceedings of the 2008 International Conference on Artificial Intelligence, ICAI 2008 and Proceedings of the 2008 International Conference on Machine Learning; Models, Technologies and Applications
Recommended Citation
Yeasin, M., Russomanno, D., Sorower, S., Smith, M., & Shaik, J. (2008). Robust classification of objects using a sparse detector sensor. Proceedings of the 2008 International Conference on Artificial Intelligence, ICAI 2008 and Proceedings of the 2008 International Conference on Machine Learning; Models, Technologies and Applications, 742-748. Retrieved from https://digitalcommons.memphis.edu/facpubs/14174
