Deep learning for NeuroImaging in Python.

Note

This page is a reference documentation. It only explains the class signature, and not how to use it. Please refer to the gallery for the big picture.

class nidl.callbacks.model_probing.KNeighborsRegressorCVCallback(train_dataloader: DataLoader, test_dataloader: DataLoader, probe_name: str | None = None, n_neighbors: tuple[int] = (2, 5, 10), cv: int = 5, n_jobs: int | None = None, scoring: str = 'r2', **kwargs)[source]¶

Bases: ModelProbing

Perform KNN regression on top of an embedding model.

Concretely, this callback:

Embeds the input data through the torch model.
Performs n-fold cross-validation to find the best n_neighbors.
Logs the main regression metrics by regressor and averaged, including:
- mean absolute error
- median absolute error
- root mean squared error
- mean squared error
- R² score
- Pearson correlation coefficient
- explained variance score

Parameters:

train_dataloader : torch.utils.data.DataLoader

Training dataloader yielding batches in the form (X, y) for further embedding and training of the KNN probe.

test_dataloader : torch.utils.data.DataLoader

Test dataloader yielding batches in the form (X, y) for further embedding and test of the KNN probe.

probe_name : str or None, default=None

Name of the probe displayed when logging the results. It will appear as <probe_name>/<metric_name> for each metric. If None, only <metric_name> is displayed.

n_neighbors : tuple of int, default=(2, 5, 10)

Arrays of n_neighbors values to try in CV. It corresponds to the number of neighbors to use by the KNN on the training set.

cv : int or cross-validation generator, default=5

How many folds to use for cross-validating the alpha regularization strength in the Ridge regression.

n_jobs : int or None, default=None

Number of jobs to run in parallel. None means 1 unless in a joblib.parallel_backend context. -1 means using all processors.

scoring : str in {“r2”, “neg_mean_absolute_error”,

“neg_mean_squared_error”, …}, default=”r2” Which scoring function to use to cross-validate the n_neighbors hyper-parameter. For a complete list of scoring options, check https://scikit-learn.org/1.4/modules/model_evaluation.html#scoring

kwargs : dict

Additional keyword arguments to pass to the ModelProbing constructor (e.g. every_n_train_epochs, every_n_val_epochs, prog_bar, …).

fit(X, y)[source]¶

Fit the k-nearest neighbors regressor from the training dataset and log the regression metrics.

Parameters:

X : {array-like, sparse matrix} of shape (n_samples, n_features)

Training data.

y : {array-like, sparse matrix} of shape (n_samples,) or (n_samples, n_outputs)

Target values.

log_metrics(pl_module, y_pred, y_true)[source]¶: Log the metrics given the predictions and the true labels.

predict(X)[source]¶: Predict the probe on new data X.