Added latent regularization control to deepof.data.coordinates.deep_unsupervised_embedding()

deepof/train_model.py

@@ -183,14 +183,6 @@ parser.add_argument(
     type=float,
     default=0.99,
 )
-parser.add_argument(
-    "--stability-check",
-    "-s",
-    help="Sets the number of times that the model is trained and initialised. "
-    "If greater than 1 (the default), saves the cluster assignments to a dataframe on disk",
-    type=int,
-    default=1,
-)
 parser.add_argument("--train-path", "-tp", help="set training set path", type=str)
 parser.add_argument(
     "--val-num",
@@ -243,7 +235,6 @@ output_path = os.path.join(args.output_path)
 overlap_loss = args.overlap_loss
 pheno_class = float(args.phenotype_classifier)
 predictor = float(args.predictor)
-runs = args.stability_check
 smooth_alpha = args.smooth_alpha
 train_path = os.path.abspath(args.train_path)
 tune = args.hyperparameter_tuning
@@ -287,7 +278,7 @@ project_coords = project(
     animal_ids=tuple([animal_id]),
     arena="circular",
     arena_dims=tuple([arena_dims]),
-    enable_iterative_imputation=True,
+    enable_iterative_imputation=False,
     exclude_bodyparts=exclude_bodyparts,
     exp_conditions=treatment_dict,
     path=train_path,
@@ -357,6 +348,10 @@ print("Done!")
 # as many times as specified by runs
 if not tune:
 
+    print(latent_reg)
+    print(("categorical" in latent_reg))
+    print(("variance" in latent_reg))
+
     trained_models = project_coords.deep_unsupervised_embedding(
         (X_train, y_train, X_val, y_val),
         batch_size=batch_size,

deepof/train_utils.py

@@ -430,7 +430,7 @@ def autoencoder_fitting(
             )
 
             if save_weights:
-                ae.save_weights("{}_final_weights.h5".format(run_ID))
+                ae.save_weights("{}{}_final_weights.h5".format(output_path, run_ID))
 
             if log_hparams:
                 # Logparams to tensorboard

deepof_experiments.smk

@@ -99,9 +99,8 @@ rule latent_regularization_experiments:
         "--encoding-size {wildcards.encs} "
         "--batch-size 256 "
         "--window-size 24 "
-        "--window-step 6 "
+        "--window-step 12 "
         "--exclude-bodyparts Tail_base,Tail_1,Tail_2,Tail_tip "
-        "--stability-check 3 "
         "--output-path {outpath}latent_regularization_experiments"
 #
 #

tests/test_build_hypermodels.py

@@ -35,7 +35,7 @@ def test_SEQ_2_SEQ_AE_hypermodel_build(input_shape):
 @given(
     encoding_size=st.integers(min_value=2, max_value=16),
     loss=st.one_of(st.just("ELBO"), st.just("MMD"), st.just("ELBO+MMD")),
-    number_of_components=st.integers(min_value=1, max_value=5),
+    noumber_of_components=st.integers(min_value=1, max_value=5),
 )
 def test_SEQ_2_SEQ_GMVAE_hypermodel_build(
     encoding_size,

tests/test_train_utils.py

@@ -75,7 +75,16 @@ def test_get_callbacks(
     loss,
 ):
     runID, tbc, cycle1c, cpc = deepof.train_utils.get_callbacks(
-        X_train, batch_size, True, variational, pheno_class, predictor, loss, None
+        X_train,
+        batch_size,
+        True,
+        variational,
+        pheno_class,
+        predictor,
+        loss,
+        True,
+        True,
+        None,
     )
     assert type(runID) == str
     assert type(tbc) == tf.keras.callbacks.TensorBoard