LinearRFNN

class hypercoil.nn.confound.LinearRFNN(model_dim: int, num_columns: int, basis_functions: Sequence[Callable[[Tensor], Tensor]] | None = None, num_response_functions: int = 10, response_duration: int = 9, leak: float = 0.001, *, key: jax.random.PRNGKey)

Model selection as a linear combination, with convolutional model augmentation.

Methods

__call__(x, *[, key])

Call self as a function.