The flexibility of these methods is often limited bytherequirement thatthecutsbeaxisaligned.

Moreoever, we develop estimation theory providing tight control on the error of any estimated trajectory, indicating that the representations could even be used in quite noise-sensitive follow-on analyses.

Also, forsimplicity, assume t =1 .

In Figure 1 (left) weseethat Gaussian posteriors (intheabsenceofdata) optimizedwith Eq. (1) convergetotheirpriorN(w|0,1)[7,68]. In CMV -MF-VI, shown in (b.), prunedunitsconvergetospiked Gaussians.

Asaconsequence, removing theerror P would reduce theloss more than removing the error N. Moreover, it is clear that this difference in error weighing increases withthelevelofimbalance between theclasses.