navlie.batch.losses.CauchyLoss

class navlie.batch.losses.CauchyLoss(c: float = 1.0)

Bases: LossFunction

loss(e: float) → float

Cauchy loss function.

The form here is taken from “MacTavish, Barfoot - At All Costs.”

Parameters:

e (float) – Residual

Returns:

Cost evaluated at a given error.

Return type:

float

weight(e: float) → float

Cauchy weight function.

Parameters:

e (float) – Residual

Returns:

Robust weight

Return type:

float

previous

navlie.batch.losses

next

navlie.batch.losses.L2Loss