eegproc.deep_learning.training_outputs module

class eegproc.deep_learning.training_outputs.CompactEpochLogger(fold_number=None, context=None)[source]

Bases: Callback

Print one consolidated training-diagnostics row per epoch.

Keras’ built-in verbose=2 logger places every metric on one very long line. This callback retains the readable metric/class/loss groupings without emitting several separate rows and leaves history.history unchanged.

Parameters:

  • fold_number (int | None, default: None)

  • context (str | None, default: None)

on_epoch_end(epoch, logs=None)[source]

Called at the end of an epoch.

Subclasses should override for any actions to run. This function should only be called during TRAIN mode.

Parameters:

  • epoch (int) – Integer, index of epoch.

  • logs (dict | None, default: None) – Dict, metric results for this training epoch, and for the validation epoch if validation is performed. Validation result keys are prefixed with val_. For training epoch, the values of the Model’s metrics are returned. Example: {‘loss’: 0.2, ‘accuracy’: 0.7}.

Return type:

None

on_train_begin(logs=None)[source]

Called at the beginning of training.

Subclasses should override for any actions to run.

Parameters:

logs (dict | None, default: None) – Dict. Currently no data is passed to this argument for this method but that may change in the future.

Return type:

None

on_train_end(logs=None)[source]

Called at the end of training.

Subclasses should override for any actions to run.

Parameters:

logs (dict | None, default: None) – Dict. Currently the output of the last call to on_epoch_end() is passed to this argument for this method but that may change in the future.

Return type:

None

class eegproc.deep_learning.training_outputs.HeldOutUserOracleMetrics(X_target, y_target, subject_ids_target, trial_ids_target, target_subject, evaluation_level, batch_size=None, decision_threshold=0.5, diagnostic_thresholds=None, ece_bins=15, calibration_shots=None, calibration_fold=None)[source]

Bases: HeldOutUserOracleMetrics

Emit one complete reporting-only oracle-prediction row per epoch.

The parent callback previously printed decoder and classification values on separate rows. This replacement preserves its initialization contract and reporting-only history while computing the complete classification suite, probability log loss, and both branch-specific reconstruction suites in one memory-bounded pass per output type.

Parameters:

  • X_target (ndarray)

  • y_target (ndarray)

  • subject_ids_target (ndarray)

  • trial_ids_target (ndarray)

  • evaluation_level (Literal['window', 'trial'])

  • batch_size (int | None, default: None)

  • decision_threshold (float, default: 0.5)

  • diagnostic_thresholds (list[float] | tuple[float, ...] | None, default: None)

  • ece_bins (int, default: 15)

  • calibration_shots (int | None, default: None)

  • calibration_fold (int | None, default: None)

on_epoch_end(epoch, logs=None)[source]

Called at the end of an epoch.

Subclasses should override for any actions to run. This function should only be called during TRAIN mode.

Parameters:

  • epoch (int) – Integer, index of epoch.

  • logs (dict | None, default: None) – Dict, metric results for this training epoch, and for the validation epoch if validation is performed. Validation result keys are prefixed with val_. For training epoch, the values of the Model’s metrics are returned. Example: {‘loss’: 0.2, ‘accuracy’: 0.7}.

Return type:

None

on_train_begin(logs=None)[source]

Called at the beginning of training.

Subclasses should override for any actions to run.

Parameters:

logs (dict | None, default: None) – Dict. Currently no data is passed to this argument for this method but that may change in the future.

Return type:

None

on_train_end(logs=None)[source]

Called at the end of training.

Subclasses should override for any actions to run.

Parameters:

logs (dict | None, default: None) – Dict. Currently the output of the last call to on_epoch_end() is passed to this argument for this method but that may change in the future.

Return type:

None

class eegproc.deep_learning.training_outputs.PredictionDiagnostics(X_train, y_train, X_val=None, y_val=None, fold_number=None, batch_size=None, every_n_epochs=1, max_samples=256, threshold_tolerance=0.01, reported_metric='accuracy', decision_threshold=0.5, decision_thresholds=None, ece_bins=15, seed=42)[source]

Bases: Callback

Print one complete deterministic training-diagnostics row per epoch.

Only a fixed, approximately class-balanced subset is evaluated, so the callback remains inexpensive relative to a full validation pass. The row combines the existing Keras epoch metrics/losses with exact probability metrics on that fixed subset, including ROC-AUC and macro F1. Decoder values come from the epoch-wide Keras logs when available; a memory-bounded inference fallback is used only for models that expose a decoder without tracking its metrics.

Parameters:

  • X_train (ndarray)

  • y_train (ndarray)

  • X_val (ndarray | None, default: None)

  • y_val (ndarray | None, default: None)

  • fold_number (int | None, default: None)

  • batch_size (int | None, default: None)

  • every_n_epochs (int, default: 1)

  • max_samples (int, default: 256)

  • threshold_tolerance (float, default: 0.01)

  • reported_metric (str, default: 'accuracy')

  • decision_threshold (float, default: 0.5)

  • decision_thresholds (Optional[Sequence[float]], default: None)

  • ece_bins (int, default: 15)

  • seed (int | None, default: 42)

on_epoch_end(epoch, logs=None)[source]

Called at the end of an epoch.

Subclasses should override for any actions to run. This function should only be called during TRAIN mode.

Parameters:

  • epoch (int) – Integer, index of epoch.

  • logs (dict | None, default: None) – Dict, metric results for this training epoch, and for the validation epoch if validation is performed. Validation result keys are prefixed with val_. For training epoch, the values of the Model’s metrics are returned. Example: {‘loss’: 0.2, ‘accuracy’: 0.7}.

Return type:

None

on_train_begin(logs=None)[source]

Called at the beginning of training.

Subclasses should override for any actions to run.

Parameters:

logs (dict | None, default: None) – Dict. Currently no data is passed to this argument for this method but that may change in the future.

Return type:

None

on_train_end(logs=None)[source]

Called at the end of training.

Subclasses should override for any actions to run.

Parameters:

logs (dict | None, default: None) – Dict. Currently the output of the last call to on_epoch_end() is passed to this argument for this method but that may change in the future.

Return type:

None