VaiPhy: aVariationalInferenceBasedAlgorithmfor Phylogeny Appendix

Neural Information Processing Systems 

Hence, during the training of VaiPhy, we used a maximum likelihood heuristic toupdate thebranch lengths givenatree topology. The branch lengths of the NJ tree are optimized with the same software. The optimized branch lengths are used as the initial set of lengths fore E(τ). In all of the figures, the left column is the current state ofτ, the middle column is two trees that are compared, and the right column is the selected tree. Solid lines indicate the edges inτ, and bold green lines are accepted edges (edges inM).

Similar Docs  Excel Report  more

TitleSimilaritySource
None found