・(事後分布が)推定困難な場合でも機能する How can we perform efficient inference and learning in directed probabilistic models, in the presence of continuous latent variables with intractable posterior distributions, and large datasets? We introduce a stochastic variational inference and learning algorithm that scales to large datasets and, under some mild differentiablity conditions, even works in the intractable case.
推定困難な事後分布であっても、効率的に推論可能 Our contributions is two-fold. First, we show that a reparameterization of the variational lower bound yields a lower bound estimator that can be straightforwardly optimized using standard stochastic gradient methods. Second, we show that for i.i.d. datasets with continuous latent variables per datapoint, posterior inference can be made especially efficient by fitting an approximate inference model (also called a recognition model) to the intractable posterior using the proposed lower bound estimator. Theoretical advantages are refleted in experimental results.
() , Gaussian case ガウス分布(正規分布)の場合のKLダイバージェンスの計算 C MLP’s as probablistic encoders and decoders MLP(多層パーセプトロン)のエンコーダ及びデコーダの定義 D Marginal likelihood estimator 周辺尤度の推定量 E Monte Carlo EM モンテカルロEMアルゴリズム F Full VB パラメータと潜在変数の両方を変分推論する場合の数式・手順
Auto-encoding variational bayes. In International Conference on Learning Representations, 2014. [2] Ha, D. and Shumidhuber, J. Recurrent World Models Facilitate Policy Evolution. NeurIPS 2018. ・書籍 [1] 変分ベイズ学習 中島伸一著(講談社) [2] ベイズ推論による機械学習入門 須山敦志著(講談社)