feat: remove output conversions (#792)

Closes partially #732

### Summary of Changes

Output conversions are the exact inversion of the input conversion, so
there is no need to specify them again. Now, a neural network only takes
an input conversion and a list of layers. This also gets rid of several
errors that could occur if input and output conversions did not fit
together.

In a later PR, the input conversion will also be removed, since they
mirror datasets.

---------

Co-authored-by: megalinter-bot <129584137+megalinter-bot@users.noreply.github.com>

src/safeds/data/labeled/containers/_time_series_dataset.py

@@ -19,9 +19,9 @@
 
 class TimeSeriesDataset:
     """
-    A time series dataset maps feature and time columns to a target column. Not like the TabularDataset a TimeSeries needs to contain one target and one time column, but can have empty features.
+    A time series dataset maps feature and time columns to a target column.
 
-    Create a time series dataset from a mapping of column names to their values.
+    Unlike a TabularDataset, a TimeSeries needs to contain one target and one time column, but can have empty features.
 
     Parameters
     ----------

src/safeds/data/tabular/containers/_table.py

@@ -551,8 +551,6 @@ def get_column(self, name: str) -> Column:
         """
         Get a column from the table.
 
-        **Note:** This operation must fully load the data into memory, which can be expensive.
-
         Parameters
         ----------
         name:
@@ -584,7 +582,9 @@ def get_column(self, name: str) -> Column:
         +-----+
         """
         _check_columns_exist(self, name)
-        return Column._from_polars_series(self._data_frame.get_column(name))
+        return Column._from_polars_series(
+            self._lazy_frame.select(name).collect().get_column(name),
+        )
 
     def get_column_type(self, name: str) -> DataType:
         """

src/safeds/exceptions/_ml.py

@@ -76,7 +76,8 @@ def __init__(self) -> None:
 class InputSizeError(Exception):
     """Raised when the amount of features being passed to a network does not match with its input size."""
 
-    def __init__(self, data_size: int | ModelImageSize, input_layer_size: int | ModelImageSize) -> None:
+    def __init__(self, data_size: int | ModelImageSize, input_layer_size: int | ModelImageSize | None) -> None:
+        # TODO: remove input_layer_size type None again
         super().__init__(
             f"The data size being passed to the network({data_size}) does not match with its input size({input_layer_size}). Consider changing the data size of the model or reformatting the data.",
         )

src/safeds/ml/nn/converters/__init__.py

@@ -6,43 +6,29 @@
 
 if TYPE_CHECKING:
     from ._input_converter import InputConversion
-    from ._input_converter_image import InputConversionImage
+    from ._input_converter_image_to_column import InputConversionImageToColumn
+    from ._input_converter_image_to_image import InputConversionImageToImage
+    from ._input_converter_image_to_table import InputConversionImageToTable
     from ._input_converter_table import InputConversionTable
     from ._input_converter_time_series import InputConversionTimeSeries
-    from ._output_converter import OutputConversion
-    from ._output_converter_image import (
-        OutputConversionImageToColumn,
-        OutputConversionImageToImage,
-        OutputConversionImageToTable,
-    )
-    from ._output_converter_table import OutputConversionTable
-    from ._output_converter_time_series import OutputConversionTimeSeries
 
 apipkg.initpkg(
     __name__,
     {
         "InputConversion": "._input_converter:InputConversion",
-        "InputConversionImage": "._input_converter_image:InputConversionImage",
+        "InputConversionImageToColumn": "._input_converter_image_to_column:InputConversionImageToColumn",
+        "InputConversionImageToImage": "._input_converter_image_to_image:InputConversionImageToImage",
+        "InputConversionImageToTable": "._input_converter_image_to_table:InputConversionImageToTable",
         "InputConversionTable": "._input_converter_table:InputConversionTable",
         "InputConversionTimeSeries": "._input_converter_time_series:InputConversionTimeSeries",
-        "OutputConversion": "._output_converter:OutputConversion",
-        "OutputConversionImageToColumn": "._output_converter_image:OutputConversionImageToColumn",
-        "OutputConversionImageToImage": "._output_converter_image:OutputConversionImageToImage",
-        "OutputConversionImageToTable": "._output_converter_image:OutputConversionImageToTable",
-        "OutputConversionTable": "._output_converter_table:OutputConversionTable",
-        "OutputConversionTimeSeries": "._output_converter_time_series:OutputConversionTimeSeries",
     },
 )
 
 __all__ = [
     "InputConversion",
-    "InputConversionImage",
+    "InputConversionImageToColumn",
+    "InputConversionImageToImage",
+    "InputConversionImageToTable",
     "InputConversionTable",
     "InputConversionTimeSeries",
-    "OutputConversion",
-    "OutputConversionImageToColumn",
-    "OutputConversionImageToImage",
-    "OutputConversionImageToTable",
-    "OutputConversionTable",
-    "OutputConversionTimeSeries",
 ]

src/safeds/ml/nn/converters/_input_converter.py

@@ -8,6 +8,7 @@
 from safeds.data.tabular.containers import Table
 
 if TYPE_CHECKING:
+    from torch import Tensor
     from torch.utils.data import DataLoader
 
     from safeds.data.image.containers._single_size_image_list import _SingleSizeImageList
@@ -22,30 +23,27 @@ class InputConversion(Generic[FT, PT], ABC):
 
     @property
     @abstractmethod
-    def _data_size(self) -> int | ModelImageSize:
-        pass  # pragma: no cover
+    def _data_size(self) -> int | ModelImageSize: ...
 
     @abstractmethod
     def _data_conversion_fit(
         self,
         input_data: FT,
         batch_size: int,
         num_of_classes: int = 1,
-    ) -> DataLoader | ImageDataset:
-        pass  # pragma: no cover
+    ) -> DataLoader | ImageDataset: ...
 
     @abstractmethod
-    def _data_conversion_predict(self, input_data: PT, batch_size: int) -> DataLoader | _SingleSizeImageList:
-        pass  # pragma: no cover
+    def _data_conversion_predict(self, input_data: PT, batch_size: int) -> DataLoader | _SingleSizeImageList: ...
 
     @abstractmethod
-    def _is_fit_data_valid(self, input_data: FT) -> bool:
-        pass  # pragma: no cover
+    def _data_conversion_output(self, input_data: PT, output_data: Tensor, **kwargs: Any) -> FT: ...
 
     @abstractmethod
-    def _is_predict_data_valid(self, input_data: PT) -> bool:
-        pass  # pragma: no cover
+    def _is_fit_data_valid(self, input_data: FT) -> bool: ...
 
     @abstractmethod
-    def _get_output_configuration(self) -> dict[str, Any]:
-        pass  # pragma: no cover
+    def _is_predict_data_valid(self, input_data: PT) -> bool: ...
+
+    @abstractmethod
+    def _get_output_configuration(self) -> dict[str, Any]: ...

src/safeds/ml/nn/converters/_input_converter_image.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import sys
+from abc import ABC
 from typing import TYPE_CHECKING, Any
 
 from safeds._utils import _structural_hash
@@ -16,25 +17,57 @@
     from safeds.ml.nn.typing import ModelImageSize
 
 
-class InputConversionImage(InputConversion[ImageDataset, ImageList]):
-    """The input conversion for a neural network, defines the input parameters for the neural network."""
+class _InputConversionImage(InputConversion[ImageDataset, ImageList], ABC):
+    """
+    The input conversion for a neural network, defines the input parameters for the neural network.
+
+    Parameters
+    ----------
+    image_size:
+        the size of the input images
+    """
 
     def __init__(self, image_size: ModelImageSize) -> None:
-        """
-        Define the input parameters for the neural network in the input conversion.
-
-        Parameters
-        ----------
-        image_size:
-            the size of the input images
-        """
         self._input_size = image_size
         self._output_size: ModelImageSize | int | None = None
         self._one_hot_encoder: OneHotEncoder | None = None
         self._column_name: str | None = None
         self._column_names: list[str] | None = None
         self._output_type: type | None = None
 
+    def __hash__(self) -> int:
+        return _structural_hash(
+            self.__class__.__name__,
+            self._input_size,
+            self._output_size,
+            self._one_hot_encoder,
+            self._column_name,
+            self._column_names,
+            self._output_type,
+        )
+
+    def __eq__(self, other: object) -> bool:
+        if not isinstance(other, type(self)):
+            return NotImplemented
+        return (self is other) or (
+            self._input_size == other._input_size
+            and self._output_size == other._output_size
+            and self._one_hot_encoder == other._one_hot_encoder
+            and self._column_name == other._column_name
+            and self._column_names == other._column_names
+            and self._output_type == other._output_type
+        )
+
+    def __sizeof__(self) -> int:
+        return (
+            sys.getsizeof(self._input_size)
+            + sys.getsizeof(self._output_size)
+            + sys.getsizeof(self._one_hot_encoder)
+            + sys.getsizeof(self._column_name)
+            + sys.getsizeof(self._column_names)
+            + sys.getsizeof(self._output_type)
+        )
+
     @property
     def _data_size(self) -> ModelImageSize:
         return self._input_size
@@ -81,64 +114,3 @@ def _get_output_configuration(self) -> dict[str, Any]:
             "column_name": self._column_name,
             "one_hot_encoder": self._one_hot_encoder,
         }
-
-    def __hash__(self) -> int:
-        """
-        Return a deterministic hash value for this InputConversionImage.
-
-        Returns
-        -------
-        hash:
-            the hash value
-        """
-        return _structural_hash(
-            self._input_size,
-            self._output_size,
-            self._one_hot_encoder,
-            self._column_name,
-            self._column_names,
-            self._output_type,
-        )
-
-    def __eq__(self, other: object) -> bool:
-        """
-        Compare two InputConversionImage instances.
-
-        Parameters
-        ----------
-        other:
-            The InputConversionImage instance to compare to.
-
-        Returns
-        -------
-        equals:
-            Whether the instances are the same.
-        """
-        if not isinstance(other, InputConversionImage):
-            return NotImplemented
-        return (self is other) or (
-            self._input_size == other._input_size
-            and self._output_size == other._output_size
-            and self._one_hot_encoder == other._one_hot_encoder
-            and self._column_name == other._column_name
-            and self._column_names == other._column_names
-            and self._output_type == other._output_type
-        )
-
-    def __sizeof__(self) -> int:
-        """
-        Return the complete size of this object.
-
-        Returns
-        -------
-        size:
-            Size of this object in bytes.
-        """
-        return (
-            sys.getsizeof(self._input_size)
-            + sys.getsizeof(self._output_size)
-            + sys.getsizeof(self._one_hot_encoder)
-            + sys.getsizeof(self._column_name)
-            + sys.getsizeof(self._column_names)
-            + sys.getsizeof(self._output_type)
-        )

src/safeds/ml/nn/converters/_input_converter_image_to_column.py

@@ -0,0 +1,55 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Any
+
+from safeds._config import _init_default_device
+from safeds.data.image.containers._single_size_image_list import _SingleSizeImageList
+from safeds.data.labeled.containers import ImageDataset
+from safeds.data.labeled.containers._image_dataset import _ColumnAsTensor
+from safeds.data.tabular.containers import Column
+from safeds.data.tabular.transformation import OneHotEncoder
+
+from ._input_converter_image import _InputConversionImage
+
+if TYPE_CHECKING:
+    from torch import Tensor
+
+    from safeds.data.image.containers import ImageList
+
+
+class InputConversionImageToColumn(_InputConversionImage):
+    def _data_conversion_output(
+        self,
+        input_data: ImageList,
+        output_data: Tensor,
+        **kwargs: Any,
+    ) -> ImageDataset[Column]:
+        import torch
+
+        _init_default_device()
+
+        if not isinstance(input_data, _SingleSizeImageList):
+            raise ValueError("The given input ImageList contains images of different sizes.")  # noqa: TRY004
+        if "column_name" not in kwargs or not isinstance(kwargs.get("column_name"), str):
+            raise ValueError(
+                "The column_name is not set. The data can only be converted if the column_name is provided as `str` in the kwargs.",
+            )
+        if "one_hot_encoder" not in kwargs or not isinstance(kwargs.get("one_hot_encoder"), OneHotEncoder):
+            raise ValueError(
+                "The one_hot_encoder is not set. The data can only be converted if the one_hot_encoder is provided as `OneHotEncoder` in the kwargs.",
+            )
+        one_hot_encoder: OneHotEncoder = kwargs["one_hot_encoder"]
+        column_name: str = kwargs["column_name"]
+
+        output = torch.zeros(len(input_data), len(one_hot_encoder._get_names_of_added_columns()))
+        output[torch.arange(len(input_data)), output_data] = 1
+
+        im_dataset: ImageDataset[Column] = ImageDataset[Column].__new__(ImageDataset)
+        im_dataset._output = _ColumnAsTensor._from_tensor(output, column_name, one_hot_encoder)
+        im_dataset._shuffle_tensor_indices = torch.LongTensor(list(range(len(input_data))))
+        im_dataset._shuffle_after_epoch = False
+        im_dataset._batch_size = 1
+        im_dataset._next_batch_index = 0
+        im_dataset._input_size = input_data.sizes[0]
+        im_dataset._input = input_data
+        return im_dataset

src/safeds/ml/nn/converters/_input_converter_image_to_image.py

@@ -0,0 +1,36 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Any
+
+from safeds._config import _init_default_device
+from safeds.data.image.containers import ImageList
+from safeds.data.image.containers._single_size_image_list import _SingleSizeImageList
+from safeds.data.labeled.containers import ImageDataset
+
+from ._input_converter_image import _InputConversionImage
+
+if TYPE_CHECKING:
+    from torch import Tensor
+
+
+class InputConversionImageToImage(_InputConversionImage):
+    def _data_conversion_output(
+        self,
+        input_data: ImageList,
+        output_data: Tensor,
+        **_kwargs: Any,
+    ) -> ImageDataset[ImageList]:
+        import torch
+
+        _init_default_device()
+
+        if not isinstance(input_data, _SingleSizeImageList):
+            raise ValueError("The given input ImageList contains images of different sizes.")  # noqa: TRY004
+
+        return ImageDataset[ImageList](
+            input_data,
+            _SingleSizeImageList._create_from_tensor(
+                (output_data * 255).to(torch.uint8),
+                list(range(output_data.size(dim=0))),
+            ),
+        )