PulseAugur
EN
LIVE 09:07:36

New framework enhances normalizing flows with stable global weighting

By PulseAugur Editorial · [2 sources] ·

Researchers have developed AMF-VI-sEMA, a novel two-stage framework for normalizing flows designed to improve approximate inference. This method uses a stable global weighting mechanism based on a Simplex Exponential Moving Average (sEMA) update. The framework trains multiple expert architectures independently in the first stage and then learns global mixture weights in the second stage, avoiding component collapse and computational overhead. AI

IMPACT This research introduces a novel approach to approximate inference in normalizing flows, potentially improving generalization across diverse posterior geometries.

RANK_REASON The cluster contains an academic paper detailing a new method for normalizing flows.

Read on arXiv stat.ML →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

New framework enhances normalizing flows with stable global weighting

COVERAGE [2]

  1. arXiv stat.ML TIER_1 English(EN) · Benjamin Wiriyapong, Oktay Karakus, Can Eyupoglu, Kirill Sidorov ·

    Stable Global Weighting of Flow Mixtures using Simplex Exponential Moving Average

    arXiv:2607.03809v1 Announce Type: cross Abstract: Normalising flows provide a powerful variational family for approximate inference, yet individual architectures often fail to generalise across heterogeneous posterior geometries. We revisit mixture-based flow formulations and int…

  2. arXiv stat.ML TIER_1 English(EN) · Kirill Sidorov ·

    Stable Global Weighting of Flow Mixtures using Simplex Exponential Moving Average

    Normalising flows provide a powerful variational family for approximate inference, yet individual architectures often fail to generalise across heterogeneous posterior geometries. We revisit mixture-based flow formulations and introduce \emph{AMF\mbox{-}VI\mbox{-}sEMA}, a two-sta…

RELATED ENTITIES

RELATED TOPICS