clarena.cl_datasets.permuted_SEMEION

  1r"""
  2The submodule in `cl_datasets` for Permuted SEMEION dataset.
  3"""
  4
  5__all__ = ["PermutedSEMEION"]
  6
  7import logging
  8from typing import Callable
  9
 10import torch
 11from torch.utils.data import Dataset, random_split
 12from torchvision.datasets import SEMEION
 13from torchvision.transforms import transforms
 14
 15from clarena.cl_datasets import CLPermutedDataset
 16
 17# always get logger for built-in logging in each module
 18pylogger = logging.getLogger(__name__)
 19
 20
 21class PermutedSEMEION(CLPermutedDataset):
 22    r"""Permuted SEMEION dataset. The [SEMEION dataset](https://archive.ics.uci.edu/dataset/178/semeion+handwritten+digit) is a collection of handwritten digits. It consists of 1,593 handwritten digit images (10 classes), each 16x16 grayscale image."""
 23
 24    original_dataset_python_class: type[Dataset] = SEMEION
 25    r"""The original dataset class."""
 26
 27    def __init__(
 28        self,
 29        root: str,
 30        num_tasks: int,
 31        test_percentage: float,
 32        validation_percentage: float,
 33        batch_size: int | dict[int, int] = 1,
 34        num_workers: int | dict[int, int] = 0,
 35        custom_transforms: (
 36            Callable
 37            | transforms.Compose
 38            | None
 39            | dict[int, Callable | transforms.Compose | None]
 40        ) = None,
 41        repeat_channels: int | None | dict[int, int | None] = None,
 42        to_tensor: bool | dict[int, bool] = True,
 43        resize: tuple[int, int] | None | dict[int, tuple[int, int] | None] = None,
 44        permutation_mode: str = "first_channel_only",
 45        permutation_seeds: dict[int, int] | None = None,
 46    ) -> None:
 47        r"""
 48        **Args:**
 49        - **root** (`str`): the root directory where the original SEMEION data 'SEMEION/' live.
 50        - **num_tasks** (`int`): the maximum number of tasks supported by the CL dataset. This decides the valid task IDs from 1 to `num_tasks`.
 51        - **test_percentage** (`float`): the percentage to randomly split some data into test data.
 52        - **validation_percentage** (`float`): the percentage to randomly split some training data into validation data.
 53        - **batch_size** (`int` | `dict[int, int]`): the batch size for train, val, and test dataloaders.
 54        If it is a dict, the keys are task IDs and the values are the batch sizes for each task. If it is an `int`, it is the same batch size for all tasks.
 55        - **num_workers** (`int` | `dict[int, int]`): the number of workers for dataloaders.
 56        If it is a dict, the keys are task IDs and the values are the number of workers for each task. If it is an `int`, it is the same number of workers for all tasks.
 57        - **custom_transforms** (`transform` or `transforms.Compose` or `None` or dict of them): the custom transforms to apply ONLY to the TRAIN dataset. Can be a single transform, composed transforms, or no transform. `ToTensor()`, normalization, permute, and so on are not included.
 58        If it is a dict, the keys are task IDs and the values are the custom transforms for each task. If it is a single transform or composed transforms, it is applied to all tasks. If it is `None`, no custom transforms are applied.
 59        - **repeat_channels** (`int` | `None` | dict of them): the number of channels to repeat for each task. Default is `None`, which means no repeat.
 60        If it is a dict, the keys are task IDs and the values are the number of channels to repeat for each task. If it is an `int`, it is the same number of channels to repeat for all tasks. If it is `None`, no repeat is applied.
 61        - **to_tensor** (`bool` | `dict[int, bool]`): whether to include the `ToTensor()` transform. Default is `True`.
 62        If it is a dict, the keys are task IDs and the values are whether to include the `ToTensor()` transform for each task. If it is a single boolean value, it is applied to all tasks.
 63        - **resize** (`tuple[int, int]` | `None` or dict of them): the size to resize the images to. Default is `None`, which means no resize.
 64        If it is a dict, the keys are task IDs and the values are the sizes to resize for each task. If it is a single tuple of two integers, it is applied to all tasks. If it is `None`, no resize is applied.
 65        - **permutation_mode** (`str`): the mode of permutation; one of:
 66            1. 'all': permute all pixels.
 67            2. 'by_channel': permute channel by channel separately. All channels are applied the same permutation order.
 68            3. 'first_channel_only': permute only the first channel.
 69        - **permutation_seeds** (`dict[int, int]` | `None`): the dict of seeds for permutation operations used to construct each task. Keys are task IDs and the values are permutation seeds for each task. Default is `None`, which creates a dict of seeds from 0 to `num_tasks`-1.
 70        """
 71
 72        super().__init__(
 73            root=root,
 74            num_tasks=num_tasks,
 75            batch_size=batch_size,
 76            num_workers=num_workers,
 77            custom_transforms=custom_transforms,
 78            repeat_channels=repeat_channels,
 79            to_tensor=to_tensor,
 80            resize=resize,
 81            permutation_mode=permutation_mode,
 82            permutation_seeds=permutation_seeds,
 83        )
 84
 85        self.test_percentage: float = test_percentage
 86        r"""The percentage to randomly split some data into test data."""
 87        self.validation_percentage: float = validation_percentage
 88        r"""The percentage to randomly split some training data into validation data."""
 89
 90    def prepare_data(self) -> None:
 91        r"""Download the original SEMEION dataset if haven't."""
 92
 93        if self.task_id != 1:
 94            return  # download all original datasets only at the beginning of first task
 95
 96        SEMEION(root=self.root_t, download=True)
 97
 98        pylogger.debug(
 99            "The original SEMEION dataset has been downloaded to %s.", self.root_t
100        )
101
102    def train_and_val_dataset(self) -> tuple[Dataset, Dataset]:
103        """Get the training and validation dataset of task `self.task_id`.
104
105        **Returns:**
106        - **train_and_val_dataset** (`tuple[Dataset, Dataset]`): the train and validation dataset of task `self.task_id`.
107        """
108        dataset_all = SEMEION(
109            root=self.root_t,
110            transform=self.train_and_val_transforms(),
111            target_transform=self.target_transform(),
112            download=False,
113        )
114
115        dataset_train_and_val, _ = random_split(
116            dataset_all,
117            lengths=[
118                1 - self.test_percentage,
119                self.test_percentage,
120            ],
121            generator=torch.Generator().manual_seed(
122                42
123            ),  # this must be set fixed to make sure the datasets across experiments are the same. Don't handle it to global seed as it might vary across experiments
124        )
125
126        return random_split(
127            dataset_train_and_val,
128            lengths=[1 - self.validation_percentage, self.validation_percentage],
129            generator=torch.Generator().manual_seed(
130                42
131            ),  # this must be set fixed to make sure the datasets across experiments are the same. Don't handle it to global seed as it might vary across experiments
132        )
133
134    def test_dataset(self) -> Dataset:
135        r"""Get the test dataset of task `self.task_id`.
136
137        **Returns:**
138        - **test_dataset** (`Dataset`): the test dataset of task `self.task_id`.
139        """
140        dataset_all = SEMEION(
141            root=self.root_t,
142            transform=self.train_and_val_transforms(),
143            target_transform=self.target_transform(),
144            download=False,
145        )
146
147        _, dataset_test = random_split(
148            dataset_all,
149            lengths=[1 - self.test_percentage, self.test_percentage],
150            generator=torch.Generator().manual_seed(42),
151        )
152
153        return dataset_test