Set default radius for latent entropy computation using an empirically derived linear equation

deepof/model_utils.py

@@ -211,12 +211,12 @@ class neighbor_cluster_purity(tf.keras.callbacks.Callback):
     """
 
     def __init__(
-        self, variational=True, validation_data=None, r=0.75, samples=10000, log_dir="."
+        self, r, variational=True, validation_data=None, samples=10000, log_dir="."
     ):
         super().__init__()
+        self.r = r
         self.variational = variational
         self.validation_data = validation_data
-        self.r = r  # Make radius default depend on encoding dimensions
         self.samples = samples
         self.log_dir = log_dir
 

deepof/train_model.py

@@ -403,7 +403,7 @@ else:
         phenotype_class=pheno_class,
         predictor=predictor,
         loss=loss,
-        logparam=None,
+        logparam=logparam,
         outpath=output_path,
     )
 

deepof/train_utils.py

@@ -75,8 +75,8 @@ def get_callbacks(
     cp: bool = False,
     reg_cat_clusters: bool = False,
     reg_cluster_variance: bool = False,
-    entropy_samples: int = 10000,
-    entropy_radius: float = 0.75,
+    entropy_samples: int = 15000,
+    entropy_radius: float = None,
     logparam: dict = None,
     outpath: str = ".",
 ) -> List[Union[Any]]:
@@ -113,7 +113,13 @@ def get_callbacks(
     )
 
     entropy = deepof.model_utils.neighbor_cluster_purity(
-        r=entropy_radius,
+        r=(
+            entropy_radius
+            if entropy_radius is not None
+            else 0.15 * logparam["encoding"]
+            - 0.18  # equation derived empirically to keep neighbor number constant.
+            # See examples/set_default_entropy_radius.ipynb for details
+        ),
         samples=entropy_samples,
         validation_data=X_val,
         log_dir=os.path.join(outpath, "metrics", run_ID),