Skip to main content
. 2017 Oct 18;11:95. doi: 10.3389/fncom.2017.00095

Figure 4.

Figure 4

(A) A Forney-style factor graph for one time step of a linear Gaussian dynamical system. (B) A 7-step sum-product message passing sequence for Kalman filtering. (C) Three additional messages (8–10) afford learning the state transition gain β from observations.