[ENH] Improve deep learning networks test coverage for parameters whi…

…ch can be list (#1851)

* Improve deep learning networks test coverage for parameters which can be list or not

* checklist : I've added myself to the list of contributors

* remove test for AE networks as they will be having their own tests for special reasons

* restore test_all_networks_functionality and add param test in test_all_networks_params

* use tag instead of the network name

* refactor test_all_networks_params to improve lisibility and skip messages

.all-contributorsrc

@@ -2447,6 +2447,15 @@
         "doc"
       ]
     },
+    {
+      "login": "Cyril-Meyer",
+      "name": "Cyril Meyer",
+      "avatar_url": "https://avatars.githubusercontent.com/u/69190238?v=4",
+      "profile": "https://github.com/Cyril-Meyer",
+      "contributions": [
+        "test"
+      ]
+    },
     {
       "login": "Moonzyyy",
       "name": "Daniele Carli",

aeon/networks/tests/test_all_networks.py

@@ -62,3 +62,75 @@ def test_all_networks_functionality(network):
             )
     else:
         pytest.skip(f"{network.__name__} not to be tested since its a base class.")
+
+
+@pytest.mark.parametrize("network", _networks)
+def test_all_networks_params(network):
+    """Test the functionality of all networks."""
+    input_shape = (100, 2)
+
+    if network.__name__ in ["BaseDeepLearningNetwork", "EncoderNetwork"]:
+        pytest.skip(f"{network.__name__} not to be tested since its a base class.")
+
+    if network._config["structure"] == "auto-encoder":
+        pytest.skip(
+            f"{network.__name__} not to be tested (AE networks have their own tests)."
+        )
+
+    if not (
+        _check_soft_dependencies(
+            network._config["python_dependencies"], severity="none"
+        )
+        and _check_python_version(network._config["python_version"], severity="none")
+    ):
+        pytest.skip(
+            f"{network.__name__} dependencies not satisfied or invalid \
+            Python version."
+        )
+
+    # check with default parameters
+    my_network = network()
+    my_network.build_network(input_shape=input_shape)
+
+    # check with list parameters
+    params = dict()
+    for attrname in [
+        "kernel_size",
+        "n_filters",
+        "avg_pool_size",
+        "activation",
+        "padding",
+        "strides",
+        "dilation_rate",
+        "use_bias",
+    ]:
+
+        # Exceptions to fix
+        if (
+            attrname in ["kernel_size", "padding"]
+            and network.__name__ == "TapNetNetwork"
+        ):
+            continue
+        # LITENetwork does not seem to work with list args
+        if network.__name__ == "LITENetwork":
+            continue
+
+        # Here we use 'None' string as default to differentiate with None values
+        attr = getattr(my_network, attrname, "None")
+        if attr != "None":
+            if attr is None:
+                attr = 3
+            elif isinstance(attr, list):
+                attr = attr[0]
+            else:
+                if network.__name__ in ["ResNetNetwork"]:
+                    attr = [attr] * my_network.n_conv_per_residual_block
+                elif network.__name__ in ["InceptionNetwork"]:
+                    attr = [attr] * my_network.depth
+                else:
+                    attr = [attr] * my_network.n_layers
+            params[attrname] = attr
+
+    if params:
+        my_network = network(**params)
+        my_network.build_network(input_shape=input_shape)