TensorFlow Models

TensorFlow/Keras models for EEG classification.

Classes:

Name	Description
TensorFlowClassifierFactory	Build Keras classifiers for the package model families.
TensorFlowConditionalGAN	Conditional GAN model for synthetic EEG training samples.
TensorFlowGANAugmenter	Train a TensorFlow conditional GAN and append generated samples.
TensorFlowTrainer	Thin TensorFlow trainer wrapper used by notebooks and examples.

Functions:

Name	Description
build_generator	Build the conditional GAN generator.
build_discriminator	Build the conditional GAN discriminator.

TensorFlowClassifierFactory

TensorFlowClassifierFactory(
    model_config: ModelConfig | None = None,
)

Build Keras classifiers for the package model families.

Parameters:

Name	Type	Description	Default
model_config	ModelConfig | None	Optional architecture configuration.	None

Methods:

Name	Description
build_cnn	Build and compile the CNN classifier.
build_lstm	Build and compile the naive LSTM classifier.
build_cnn_lstm	Build and compile the hybrid CNN-LSTM classifier.
build	Build a named TensorFlow classifier.

build_cnn

build_cnn(learning_rate: float = 0.001) -> Model

Build and compile the CNN classifier.

Parameters:

Name	Type	Description	Default
learning_rate	float	Adam optimizer learning rate.	0.001

Returns:

Type	Description
Model	Compiled Keras model.

build_lstm

build_lstm(learning_rate: float = 0.001) -> Model

Build and compile the naive LSTM classifier.

Parameters:

Name	Type	Description	Default
learning_rate	float	Adam optimizer learning rate.	0.001

Returns:

Type	Description
Model	Compiled Keras model.

build_cnn_lstm

build_cnn_lstm(learning_rate: float = 0.001) -> Model

Build and compile the hybrid CNN-LSTM classifier.

Parameters:

Name	Type	Description	Default
learning_rate	float	Adam optimizer learning rate.	0.001

Returns:

Type	Description
Model	Compiled Keras model.

build

build(name: str, learning_rate: float = 0.001) -> Model

Build a named TensorFlow classifier.

Parameters:

Name	Type	Description	Default
name	str	Classifier name: cnn, lstm, or cnn_lstm.	required
learning_rate	float	Adam optimizer learning rate.	0.001

Returns:

Type	Description
Model	Compiled Keras model.

Raises:

Type	Description
ValueError	If the classifier name is unknown.

TensorFlowConditionalGAN

TensorFlowConditionalGAN(
    discriminator: Model | None = None,
    generator: Model | None = None,
    config: ModelConfig | None = None,
)

Bases: _TensorFlowGANBase

Conditional GAN model for synthetic EEG training samples.

Parameters:

Name	Type	Description	Default
discriminator	Model | None	Optional discriminator model.	None
generator	Model | None	Optional generator model.	None
config	ModelConfig | None	Optional architecture configuration.	None

TensorFlowGANAugmenter dataclass

TensorFlowGANAugmenter(
    training: TrainingConfig,
    model_config: ModelConfig | None = None,
)

Train a TensorFlow conditional GAN and append generated samples.

The augmenter is classifier-independent: it receives prepared training arrays and returns augmented arrays that can be passed to CNN, LSTM, or CNN-LSTM classifiers.

Attributes:

Name	Type	Description
training	TrainingConfig	Training and GAN augmentation configuration.
model_config	ModelConfig | None	Optional architecture configuration.

Methods:

Name	Description
augment_training_data	Create GAN-augmented training arrays.

augment_training_data

augment_training_data(
    x_train: ndarray, y_train: ndarray
) -> GANAugmentationResult

Create GAN-augmented training arrays.

Parameters:

Name	Type	Description	Default
x_train	ndarray	Real training inputs shaped (n, time, 1, channels).	required
y_train	ndarray	One-hot real training labels.	required

Returns:

Type	Description
GANAugmentationResult	Augmented training arrays plus synthetic sample metadata.

TensorFlowTrainer dataclass

TensorFlowTrainer(training: TrainingConfig)

Thin TensorFlow trainer wrapper used by notebooks and examples.

Attributes:

Name	Type	Description
training	TrainingConfig	Training configuration controlling batch size, epochs, and dev mode.

Methods:

Name	Description
fit	Train a compiled Keras model.
evaluate	Evaluate a trained Keras model.

fit

fit(
    model: Model,
    x_train: Any,
    y_train: Any,
    x_valid: Any,
    y_valid: Any,
) -> Any

Train a compiled Keras model.

Parameters:

Name	Type	Description	Default
model	Model	Compiled classifier.	required
x_train	Any	Training inputs.	required
y_train	Any	Training one-hot labels.	required
x_valid	Any	Validation inputs.	required
y_valid	Any	Validation one-hot labels.	required

Returns:

Type	Description
Any	Keras History object.

evaluate

evaluate(
    model: Model, x_test: Any, y_test: Any
) -> dict[str, float]

Evaluate a trained Keras model.

Parameters:

Name	Type	Description	Default
model	Model	Trained classifier.	required
x_test	Any	Test inputs.	required
y_test	Any	Test one-hot labels.	required

Returns:

Type	Description
dict[str, float]	Dictionary with loss and accuracy.

build_generator

build_generator(config: ModelConfig | None = None) -> Model

Build the conditional GAN generator.

Parameters:

Name	Type	Description	Default
config	ModelConfig | None	Optional architecture configuration.	None

Returns:

Type	Description
Model	Keras generator model.

build_discriminator

build_discriminator(
    config: ModelConfig | None = None,
) -> Model

Build the conditional GAN discriminator.

Parameters:

Name	Type	Description	Default
config	ModelConfig | None	Optional architecture configuration.	None

Returns:

Type	Description
Model	Keras discriminator model.