arrow
Volume 15, Issue 4
Broad Learning System with Preprocessing to Recover the Scattering Obstacles with Far–Field Data

Weishi Yin, Hongyu Qi & Pinchao Meng

Adv. Appl. Math. Mech., 15 (2023), pp. 984-1000.

Published online: 2023-04

Export citation
  • Abstract

Based on Broad Learning System with preprocessing, the impenetrable obstacles were reconstructed. Firstly, the far-field data were preprocessed by Random Forest, and the shapes of the obstacles were classified by dividing the far-field data into different categories. Secondly, the broad learning system was employed for reconstructing the unknown scatterer. The far-field data of the scatterer were regarded as the input nodes of mapped features in the network, and all the mapped features were connected with the enhancement nodes of random weights to the output layer. Subsequently, the coefficient of the output can be obtained by the pseudoinverse. This method for the recovery of the scattering obstacles is named RF-BLS. Finally, numerical experiments revealed that the proposed method is effective, and that the training speed was significantly improved, compared with the deep learning method.

  • AMS Subject Headings

35R30, 65N21

  • Copyright

COPYRIGHT: © Global Science Press

  • Email address
  • BibTex
  • RIS
  • TXT
@Article{AAMM-15-984, author = {Yin , WeishiQi , Hongyu and Meng , Pinchao}, title = {Broad Learning System with Preprocessing to Recover the Scattering Obstacles with Far–Field Data}, journal = {Advances in Applied Mathematics and Mechanics}, year = {2023}, volume = {15}, number = {4}, pages = {984--1000}, abstract = {

Based on Broad Learning System with preprocessing, the impenetrable obstacles were reconstructed. Firstly, the far-field data were preprocessed by Random Forest, and the shapes of the obstacles were classified by dividing the far-field data into different categories. Secondly, the broad learning system was employed for reconstructing the unknown scatterer. The far-field data of the scatterer were regarded as the input nodes of mapped features in the network, and all the mapped features were connected with the enhancement nodes of random weights to the output layer. Subsequently, the coefficient of the output can be obtained by the pseudoinverse. This method for the recovery of the scattering obstacles is named RF-BLS. Finally, numerical experiments revealed that the proposed method is effective, and that the training speed was significantly improved, compared with the deep learning method.

}, issn = {2075-1354}, doi = {https://doi.org/10.4208/aamm.OA-2021-0352}, url = {http://global-sci.org/intro/article_detail/aamm/21599.html} }
TY - JOUR T1 - Broad Learning System with Preprocessing to Recover the Scattering Obstacles with Far–Field Data AU - Yin , Weishi AU - Qi , Hongyu AU - Meng , Pinchao JO - Advances in Applied Mathematics and Mechanics VL - 4 SP - 984 EP - 1000 PY - 2023 DA - 2023/04 SN - 15 DO - http://doi.org/10.4208/aamm.OA-2021-0352 UR - https://global-sci.org/intro/article_detail/aamm/21599.html KW - Inverse scattering problem, broad learning system, machine learning, random forest. AB -

Based on Broad Learning System with preprocessing, the impenetrable obstacles were reconstructed. Firstly, the far-field data were preprocessed by Random Forest, and the shapes of the obstacles were classified by dividing the far-field data into different categories. Secondly, the broad learning system was employed for reconstructing the unknown scatterer. The far-field data of the scatterer were regarded as the input nodes of mapped features in the network, and all the mapped features were connected with the enhancement nodes of random weights to the output layer. Subsequently, the coefficient of the output can be obtained by the pseudoinverse. This method for the recovery of the scattering obstacles is named RF-BLS. Finally, numerical experiments revealed that the proposed method is effective, and that the training speed was significantly improved, compared with the deep learning method.

Weishi Yin, Hongyu Qi & Pinchao Meng. (2023). Broad Learning System with Preprocessing to Recover the Scattering Obstacles with Far–Field Data. Advances in Applied Mathematics and Mechanics. 15 (4). 984-1000. doi:10.4208/aamm.OA-2021-0352
Copy to clipboard
The citation has been copied to your clipboard