[ENH] Test coverage for Resnet Network

aeon/networks/tests/test_resnet.py

@@ -0,0 +1,109 @@
+"""Tests for the ResNet Model."""
+
+import pytest
+
+from aeon.networks import ResNetNetwork
+from aeon.utils.validation._dependencies import _check_soft_dependencies
+
+
+@pytest.mark.skipif(
+    not _check_soft_dependencies(["tensorflow"], severity="none"),
+    reason="skip test if required soft dependency not available",
+)
+def test_default_initialization():
+    """Test if the network initializes with proper attributes."""
+    model = ResNetNetwork()
+    assert isinstance(
+        model, ResNetNetwork
+    ), "Model initialization failed: Incorrect type"
+    assert model.n_residual_blocks == 3, "Default residual blocks count mismatch"
+    assert (
+        model.n_conv_per_residual_block == 3
+    ), "Default convolution blocks count mismatch"
+    assert model.n_filters is None, "Default n_filters should be None"
+    assert model.kernel_size is None, "Default kernel_size should be None"
+    assert model.strides == 1, "Default strides value mismatch"
+    assert model.dilation_rate == 1, "Default dilation rate mismatch"
+    assert model.activation == "relu", "Default activation mismatch"
+    assert model.use_bias is True, "Default use_bias mismatch"
+    assert model.padding == "same", "Default padding mismatch"
+
+
+@pytest.mark.skipif(
+    not _check_soft_dependencies(["tensorflow"], severity="none"),
+    reason="skip test if required soft dependency not available",
+)
+def test_custom_initialization():
+    """Test whether custom kwargs are correctly set."""
+    model = ResNetNetwork(
+        n_residual_blocks=3,
+        n_conv_per_residual_block=3,
+        n_filters=[64, 128, 128],
+        kernel_size=[8, 5, 3],
+        activation="relu",
+        strides=1,
+        padding="same",
+    )
+    model.build_network((128, 1))
+    assert isinstance(
+        model, ResNetNetwork
+    ), "Custom initialization failed: Incorrect type"
+    assert model._n_filters == [64, 128, 128], "n_filters list mismatch"
+    assert model._kernel_size == [8, 5, 3], "kernel_size list mismatch"
+    assert model._activation == ["relu", "relu", "relu"], "activation list mismatch"
+    assert model._strides == [1, 1, 1], "strides list mismatch"
+    assert model._padding == ["same", "same", "same"], "padding list mismatch"
+
+
+@pytest.mark.skipif(
+    not _check_soft_dependencies(["tensorflow"], severity="none"),
+    reason="skip test if required soft dependency not available",
+)
+def test_invalid_initialization():
+    """Test if the network raises valid exceptions for invalid configurations."""
+    with pytest.raises(ValueError, match=".*same as number of residual blocks.*"):
+        ResNetNetwork(n_filters=[64, 128], n_residual_blocks=3).build_network((128, 1))
+
+    with pytest.raises(ValueError, match=".*same as number of convolution layers.*"):
+        ResNetNetwork(kernel_size=[8, 5], n_conv_per_residual_block=3).build_network(
+            (128, 1)
+        )
+
+    with pytest.raises(ValueError, match=".*same as number of convolution layers.*"):
+        ResNetNetwork(strides=[1, 2], n_conv_per_residual_block=3).build_network(
+            (128, 1)
+        )
+
+
+@pytest.mark.skipif(
+    not _check_soft_dependencies(["tensorflow"], severity="none"),
+    reason="skip test if required soft dependency not available",
+)
+def test_build_network():
+    """Test network building with various input shapes."""
+    model = ResNetNetwork()
+
+    input_shapes = [(128, 1), (256, 3), (512, 1)]
+    for shape in input_shapes:
+        input_layer, output_layer = model.build_network(shape)
+        assert hasattr(input_layer, "shape"), "Input layer type mismatch"
+        assert hasattr(output_layer, "shape"), "Output layer type mismatch"
+        assert input_layer.shape[1:] == shape, "Input shape mismatch"
+        assert output_layer.shape[-1] == 128, "Output layer mismatch"
+
+
+@pytest.mark.skipif(
+    not _check_soft_dependencies(["tensorflow"], severity="none"),
+    reason="skip test if required soft dependency not available",
+)
+def test_shortcut_layer():
+    """Test the shortcut layer functionality."""
+    model = ResNetNetwork()
+
+    input_shape = (128, 64)
+    input_layer, output_layer = model.build_network(input_shape)
+
+    shortcut = model._shortcut_layer(input_layer, output_layer)
+
+    assert hasattr(shortcut, "shape"), "Shortcut layer output type mismatch"
+    assert shortcut.shape[-1] == 128, "Shortcut output shape mismatch"