Group-sparsity-aware-CNN

This repository is the implementation of group sparsity-aware CNN for continuous missing data recovery of structural health monitoring as described in the paper

"Group sparsity-aware convolutional neural network for continuous missing data recovery of structural health monitoring" by Zhiyi Tang, Yuequan Bao, and Hui Li.

Packages dependencies are listed in requirements.txt. The GS-aware CNN, and the data pre- and post-processing are packed into the main function gsn(). It works as an automatic work flow once given required parameters and the data-to-recover. Two example data sets have been included in folder \simulation_El_Centro and \simulation_impulse. Run GS-aware_CNN.py with the default parameters to check the examples.