Trying to add a sparsity loss to learned coefficients for inverse problems

[sheikhu28]

As the title says, I am trying to add a sparsity loss to the learned parameters corresponding to the inverse problem in question. Any ideas on how to do this efficiently in DeepXDE?

[lululxvi]

What is sparsity loss?

[sheikhu28]

My bad, I meant something like an L1 penalty on the learned coefficients in hopes of encouraging sparsity among them.

[lululxvi]

Then use L1 regularization.