The hope is that we can design some regularization scheme that provides us the performance of neural networks while providing an interpretation that is understandable. Tree regularization does not do this, but it is one approach that uses regularization for interpretability.
The hope is that we can design some regularization scheme that provides us the performance of neural networks while providing an interpretation that is understandable. Tree regularization does not do this, but it is one approach that uses regularization for interpretability.