diff --git a/sbBrix/common/policies.py b/sbBrix/common/policies.py
index df1c1cc..66de5c8 100644
--- a/sbBrix/common/policies.py
+++ b/sbBrix/common/policies.py
@@ -580,6 +580,10 @@ class ActorCriticPolicy(BasePolicy):
             self.action_net, self.log_std = self.action_dist.proba_distribution_net(
                 latent_dim=latent_dim_pi, latent_sde_dim=latent_dim_pi, log_std_init=self.log_std_init
             )
+        elif isinstance(self.action_dist, PCA_Distribution):
+            self.action_net, self.log_std = self.action_dist.proba_distribution_net(
+                latent_dim=latent_dim_pi
+            )
         elif isinstance(self.action_dist, (CategoricalDistribution, MultiCategoricalDistribution, BernoulliDistribution)):
             self.action_net = self.action_dist.proba_distribution_net(latent_dim=latent_dim_pi)
         else: