Summary of Variational Bayesian Phylogenetic Inference with Semi-implicit Branch Length Distributions, by Tianyu Xie et al.
Variational Bayesian Phylogenetic Inference with Semi-implicit Branch Length Distributions
by Tianyu Xie, Frederick A. Matsen IV, Marc A. Suchard, Cheng Zhang
First submitted to arxiv on: 9 Aug 2024
Categories
- Main: Machine Learning (stat.ML)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper addresses the challenge of reconstructing evolutionary histories from molecular sequences using Bayesian phylogenetic inference. Markov chain Monte Carlo methods are commonly used but can be inefficient when dealing with large datasets. As an alternative, variational Bayesian phylogenetic inference (VBPI) transforms the inference problem into an optimization task. The default diagonal lognormal approximation for branch lengths in VBPI is often insufficient to capture complexity. This paper proposes a more flexible family of branch length variational posteriors based on semi-implicit hierarchical distributions using graph neural networks. The method emits permutation equivariant distributions, allowing it to handle non-Euclidean branch length spaces across different tree topologies. To optimize the proposed method, several alternative lower bounds are developed for stochastic optimization. The paper demonstrates the effectiveness of its approach over baseline methods on benchmark data examples in terms of marginal likelihood estimation and branch length posterior approximation. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research helps us understand how molecules evolved over time by analyzing their sequences. Right now, computer algorithms use something called Markov chain Monte Carlo to do this job, but it can get stuck when there are many sequences to analyze. The researchers found a new way to do this using something called variational Bayesian phylogenetic inference (VBPI). They made it better by adding more details about the branch lengths in the tree-like structure of evolutionary history. This new method works well and is faster than older methods. It’s like having a superpower for understanding how life on Earth changed over time! |
Keywords
* Artificial intelligence * Inference * Likelihood * Optimization
