SLDA

class capymoa.ocl.strategy.SLDA

Bases: BatchClassifier

Streaming Linear Discriminant Analysis.

Streaming Linear Discriminant Analysis (SLDA) [1] is a prototype classifier that incrementally accumulates the mean of each class and the mean and covariance across all classes. Note that this method does not gracefully forget and may not handle concept drift well.

[1] uses SLDA ontop of a pre-trained model to perform continual learning. See [2] and [3] for more details on incremental LDA outside of continual learning.

>>> from capymoa.ocl.strategy import SLDA
>>> from capymoa.ocl.datasets import TinySplitMNIST
>>> from capymoa.ocl.evaluation import ocl_train_eval_loop
>>> scenario = TinySplitMNIST()
>>> learner = SLDA(scenario.schema)
>>> results = ocl_train_eval_loop(
...     learner,
...     scenario.train_loaders(32),
...     scenario.test_loaders(32),
... )
>>> print(f"{results.accuracy_final*100:.1f}%")
75.5%

[1] (1,2)

Hayes, T. L., & Kanan, C. (2020). Lifelong Machine Learning with Deep Streaming Linear Discriminant Analysis. CLVision Workshop at CVPR 2020, 1–15.

[2]

Ghassabeh, Y. A., Rudzicz, F., & Moghaddam, H. A. (2015). Fast incremental LDA feature extraction. Pattern Recognition, 48(6), 1999-2012.

[3]

Linear discriminant analysis. (2025). In Wikipedia.

__init__(

schema: Schema,

pre_processor: Module = nn.Identity(),

num_features: int | None = None,

ridge: float = 1e-6,

device: device | str = torch.device('cpu'),

)

Initialize a SLDA classifier.

Parameters:

• schema – Describes the shape and type of the data.

• pre_processor – A pre-processing module to apply to the input data, defaults to an identity module.

• num_features – Number of features once pre-processed, defaults to the number of attributes in the schema.

• ridge – Ridge regularization term to avoid singular covariance matrix, defaults to 1e-6.

• device – Device to run the model on, defaults to CPU.

batch_predict(x: Tensor) → Tensor

Predict the labels for a batch of instances.

Parameters:

x – Batch of x_dtype valued feature vectors (batch_size, num_features)

Returns:

Predicted batch of y_dtype valued labels (batch_size,).

batch_predict_proba(x: Tensor) → Tensor

Predict the probabilities of the classes for a batch of instances.

Parameters:

x – Batch of x_dtype valued feature vectors (batch_size, num_features)

Returns:

Batch of x_dtype valued predicted probabilities (batch_size, num_classes).

batch_train(x: Tensor, y: Tensor) → None

Train with a batch of instances.

Parameters:

• x – Batch of x_dtype valued feature vectors (batch_size, num_features)

• y – Batch of y_dtype valued labels (batch_size,).

predict(instance: Instance) → int | None

Predict the label of an instance.

The base implementation calls predict_proba() and returns the label with the highest probability.

Parameters:

instance – The instance to predict the label for.

Returns:

The predicted label or None if the classifier is unable to make a prediction.

predict_proba(

instance: Instance,

) → ndarray[Any, dtype[float64]] | None

Calls batch_predict_proba() with a batch of size 1.

train(instance: LabeledInstance) → None

Calls batch_train() with a batch of size 1.

device: device = device(type='cpu')

Device on which the batch will be processed.

random_seed: int

The random seed for reproducibility.

When implementing a classifier ensure random number generators are seeded.

schema: Schema

The schema representing the instances.

x_dtype: dtype = torch.float32

Data type for the input features.

y_dtype: dtype = torch.int64

Data type for the target value/labels.