Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper investigates the local-mass behavior of Bayesian inference, challenging the adequacy of global objectives like KL divergence and ELBO in capturing distributional discrepancies. By introducing the Mass Index and regularised extended KL (RE-KL), the authors characterize how Bayesian updating influences local mass, revealing that power-log likelihood factors and parameter-dependent supports can significantly alter local scales. The findings include inequalities for comparing local small-ball masses, providing a theoretical foundation for understanding local mass behavior, supported by experimental illustrations.
Local mass behavior in Bayesian inference reveals that traditional global objectives may overlook critical dynamics, impacting how we understand model updates.
Global objectives, such as KL divergence and ELBO, are widely used in Bayesian inference for measuring distributional discrepancy. This paper studies their local-mass behaviour that is not directly captured by such objectives. We introduce and use two mathematical tools: (1) Mass Index for recording the polynomial and logarithmic decay scales of local mass, and (2) regularised extended KL (RE-KL), a set-localised divergence that can be formulated in the presence of singular components. Mass Indices help characterise how Bayesian updating changes local mass: (1) power-log likelihood factors shift it explicitly, and (2) parameter-dependent supports, or their smooth softenings, may change the local scale through the amount of mass that remains near the parameter value. Using local RE-KL, we prove absolute, relative, and directional inequalities for comparing local small-ball masses under the two KL directions. Together, these results provide a local theoretical account of local mass behaviour. Experiments provide controlled illustrations of the local behaviour. Code is available at https://github.com/Forsythia0604/Local-Mass-Framework.