Detecting Parameter Symmetries in Probabilistic Models
This addresses challenges in inference and interpretation for users of probabilistic models, though it is incremental as it builds on existing symmetry concepts.
The paper tackled the problem of detecting parameter symmetries in probabilistic models, which cause correlation and multimodality in posterior distributions, by introducing local symmetries and deriving algorithms for automatic detection, showing compatibility with probabilistic programming constructs and scalability to many variables.
Probabilistic models often have parameters that can be translated, scaled, permuted, or otherwise transformed without changing the model. These symmetries can lead to strong correlation and multimodality in the posterior distribution over the model's parameters, which can pose challenges both for performing inference and interpreting the results. In this work, we address the automatic detection of common problematic model symmetries. To do so, we introduce local symmetries, which cover many common cases and are amenable to automatic detection. We show how to derive algorithms to detect several broad classes of local symmetries. Our algorithms are compatible with probabilistic programming constructs such as arrays, for loops, and if statements, and they scale to models with many variables.