Added SHADE

README.md

@@ -132,6 +132,7 @@ Remove clustpy via pip to avoid ambiguities during development, e.g., when chang
 | ENRC       | [Deep Embedded Non-Redundant Clustering](https://ojs.aaai.org/index.php/AAAI/article/view/5961) | AAAI 2020                        | [Link](https://gitlab.cs.univie.ac.at/lukas/enrcpublic) (Python + PyTorch)    | [Link](https://clustpy.readthedocs.io/en/latest/clustpy.deep.html#clustpy.deep.enrc.ENRC)             |
 | IDEC       | [Improved Deep Embedded Clustering with Local Structure Preservation](https://www.ijcai.org/proceedings/2017/243) | IJCAI 2017                       | [Link](https://github.com/XifengGuo/IDEC) (Python + Keras)                    | [Link](https://clustpy.readthedocs.io/en/latest/clustpy.deep.html#clustpy.deep.dec.IDEC)              |
 | N2D        | [N2d:(not too) deep clustering via clustering the local manifold of an autoencoded embedding](https://ieeexplore.ieee.org/document/9413131) | ICPR 2021                        | [Link](https://github.com/XifengGuo/IDEC) (Python + Keras)                    | [Link](https://clustpy.readthedocs.io/en/latest/clustpy.deep.html#clustpy.deep.ddc_n2d.N2D)           |
+| SHADE | [SHADE: Deep Density-based Clustering](https://ieeexplore.ieee.org/abstract/document/10884477) | ICDM 2024 | [Link](https://github.com/pasiweber/SHADE) (Python + PyTorch) | [Link](https://clustpy.readthedocs.io/en/latest/clustpy.deep.html#clustpy.deep.shade.SHADE) |
 | VaDE       | [Variational Deep Embedding: An Unsupervised and Generative Approach to Clustering](https://www.ijcai.org/proceedings/2017/0273) | IJCAI 2017                       | [Link](https://github.com/slim1017/VaDE) (Python + Keras)                     | [Link](https://clustpy.readthedocs.io/en/latest/clustpy.deep.html#clustpy.deep.vade.VaDE)             |
 
 #### Neural Networks

clustpy/deep/__init__.py

@@ -9,6 +9,7 @@
 from .aec import AEC
 from .deepect import DeepECT
 from .den import DEN
+from .shade import SHADE
 from ._data_utils import get_dataloader, get_default_augmented_dataloaders
 from ._train_utils import get_trained_network, get_neural_network
 from ._utils import encode_batchwise, decode_batchwise, encode_decode_batchwise, predict_batchwise, detect_device, \
@@ -28,6 +29,7 @@
            'DipEncoder',
            'DeepECT',
            'DEN',
+           'SHADE',
            'get_dataloader',
            'get_default_augmented_dataloaders',
            'get_neural_network'

clustpy/deep/_abstract_deep_clustering_algo.py

@@ -101,11 +101,6 @@ def transform(self, X: np.ndarray) -> np.ndarray:
         """
         check_is_fitted(self, ["labels_", "neural_network_trained_", "n_features_in_"])
         X, _, _ = check_parameters(X, allow_size_1=True, allow_nd=self.neural_network_trained_.allow_nd_input, estimator_obj=self)
-        if X.shape[1] != self.n_features_in_:
-            raise ValueError(
-            f"X has {X.shape[1]} features, but {self.__class__.__name__} "
-            f"is expecting {self.n_features_in_} features as input."
-        )
         X_embed = self.neural_network_trained_.transform(X)
         return X_embed
 

clustpy/deep/_train_utils.py

@@ -167,7 +167,7 @@ def get_trained_network(trainloader: torch.utils.data.DataLoader = None, data: n
     neural_network = get_neural_network(input_dim, embedding_size, neural_network, neural_network_class,
                                          neural_network_params, neural_network_weights, device, random_state)
     # Start training
-    if not neural_network.fitted:
+    if not neural_network.fitted and n_epochs > 0:
         print("Neural network is not fitted yet, will be pretrained.")
         # Pretrain neural network
         optimizer_params = {"lr": 1e-3} if optimizer_params is None else optimizer_params