StatML

Time-steady States on Systems

動的システムの定常状態,線形性,移動平均について.

State Space Model & Particle Filter

State Space Model (SSM) State Space Model(SSM) is widely used in the field requiring the sequential estimation or online learning. This model is effective if you consider a system having two different variables; one completely represents the actual state but cannot be observed and the other partially represents the actual state but can be observed.

Free energy and Bayes inference

Fristonの自由エネルギー原理とELBOの等価性について

Deriving ELBO

Evidence Lower Bound (ELBO) is widely used in variational inference. Recently, according to the massive success of DeepLearning and related models, variational inference (and its technic) gains exposure in the filed of representation learning.

Counterfactual Regret Minimization

In this post, I introduce you the Counterfactual Regret Minimization (CFR Algorithm). It is mainly used for the algorithm to figure out the optimal strategy of a extensive-form game with incomplete information such as Poker and Mahjong.

EM Algorithm

​ EM algorithm is an algorithm for deriving the maximum likelihood estimator (MLE), which is generally applied to statistical methods for incomplete data. Originally, the concept of “in

Kullback-Leibler Divergence

This is a basic notebook for KL-divergence which frequently appears in many fields such as statistics and information theory.