Inverse Problems Over Probability Measure Space

Abstract: Define a forward problem as $\rho_y = G_#\rho_x$, where the probability distribution $\rho_x$ is mapped to another distribution $\rho_y$ using the forward operator $G$. In this work, we investigate the corresponding inverse problem: Given $\rho_y$, how to find $\rho_x$? Depending on whether $G$ is overdetermined or underdetermined, the solution can have drastically different behavior. In the overdetermined case, we formulate a variational problem $\min_{\rho_x} D( G_#\rho_x, \rho_y)$, and find that different choices of the metric $D$ significantly affect the quality of the reconstruction. When $D$ is set to be the Wasserstein distance, the reconstruction is the marginal distribution, while setting $D$ to be a $\phi$-divergence reconstructs the conditional distribution. In the underdetermined case, we formulate the constrained optimization $\min_{{ G_#\rho_x=\rho_y}} E[\rho_x]$. The choice of $E$ also significantly impacts the construction: setting $E$ to be the entropy gives us the piecewise constant reconstruction, while setting $E$ to be the second moment, we recover the classical least-norm solution. We also examine the formulation with regularization: $\min_{\rho_x} D( G_#\rho_x, \rho_y) + \alpha \mathsf R[\rho_x]$, and find that the entropy-entropy pair leads to a regularized solution that is defined in a piecewise manner, whereas the $W_2$-$W_2$ pair leads to a least-norm solution where $W_2$ is the 2-Wasserstein metric.