Dynamics-Aligned Representation Learning under Induced Partial Observability
A stochastic gating mechanism that induces controlled partial observability during training by sampling a spatial binary mask from a Beta--Bernoulli random field and applying it to the raw observation is proposed.