Research Article
Multi-kernel and Multi-task Learning for Radar Target Recognition
@INPROCEEDINGS{10.1007/978-3-030-67514-1_31, author={Cong Li and Xianyu Wang and Xu Yang}, title={Multi-kernel and Multi-task Learning for Radar Target Recognition}, proceedings={IoT as a Service. 6th EAI International Conference, IoTaaS 2020, Xi’an, China, November 19--20, 2020, Proceedings}, proceedings_a={IOTAAS}, year={2021}, month={1}, keywords={Multi-kernel learning (MKL) Multi-task learning (MTL) Radar target recognition Synthetic aperture radar (SAR)}, doi={10.1007/978-3-030-67514-1_31} }- Cong Li
Xianyu Wang
Xu Yang
Year: 2021
Multi-kernel and Multi-task Learning for Radar Target Recognition
IOTAAS
Springer
DOI: 10.1007/978-3-030-67514-1_31
Abstract
In this paper, a multiple kernel and multiple task learning framework (MKMTL) is proposed. To improve the interpretability of input data and adapt to different data sets, a weighted data-dependent kernel function is proposed and extended to multiple kernel functions. To fully reveal and utilize the shared information among different radar targets, multi-task learning framework is proposed. In this paper, a larger class of mixed norm penalty is adopted. It can increase the flexibility of MKMTL model. To verify the performance of the proposed model, measured MSTAR SAR public database is conducted. Experimental results demonstrate that the proposed method can effectively utilize the shared or potential information among different tasks and exhibits a better recognition performance compared with several popular existing recognition methods.
