def normalize(self, gradients: ep.Tensor, *, x: ep.Tensor,
bounds: Bounds) -> ep.Tensor:
bad_pos = ep.logical_or(
ep.logical_and(x == bounds.lower, gradients < 0),
ep.logical_and(x == bounds.upper, gradients > 0),
)
gradients = ep.where(bad_pos, ep.zeros_like(gradients), gradients)
abs_gradients = gradients.abs()
quantiles = np.quantile(flatten(abs_gradients).numpy(),
q=self.quantile,
axis=-1)
keep = abs_gradients >= atleast_kd(ep.from_numpy(gradients, quantiles),
gradients.ndim)
e = ep.where(keep, gradients.sign(), ep.zeros_like(gradients))
return normalize_lp_norms(e, p=1)
def get_perturbations(self, distances: ep.Tensor, grads: ep.Tensor) -> ep.Tensor:
return atleast_kd(distances, grads.ndim) * grads.sign()
def normalize(self, gradients: ep.Tensor, *, x: ep.Tensor,
bounds: Bounds) -> ep.Tensor:
return gradients.sign()
def normalize(self, gradients: ep.Tensor) -> ep.Tensor:
return gradients.sign()
