Prior Density Learning in Variational Bayesian Phylogenetic Parameters Inference

The advances in variational inference are providing promising paths Bayesian estimation problems. These make phylogenetic an alternative approach to Markov Chain Monte Carlo methods for approximating the posterior. However, one of main drawbacks such approaches is modelling prior through fixed distributions, which could bias posterior approximation if they distant from current data distribution. In this paper, we propose and implementation framework relax rigidity densities by learning their parameters using a gradient-based method neural network-based parameterization. We applied branch lengths evolutionary under several chain substitution models. results performed simulations show that powerful estimating model parameters. They also flexible provide better than predefined model. Finally, highlight networks improves initialization optimization density