Fast Differentiable Clipping-Aware Normalization and Rescaling

07/15/2020
by   Jonas Rauber, et al.
6

Rescaling a vector δ⃗∈ℝ^n to a desired length is a common operation in many areas such as data science and machine learning. When the rescaled perturbation ηδ⃗ is added to a starting point x⃗∈ D (where D is the data domain, e.g. D = [0, 1]^n), the resulting vector v⃗ = x⃗ + ηδ⃗ will in general not be in D. To enforce that the perturbed vector v is in D, the values of v⃗ can be clipped to D. This subsequent element-wise clipping to the data domain does however reduce the effective perturbation size and thus interferes with the rescaling of δ⃗. The optimal rescaling η to obtain a perturbation with the desired norm after the clipping can be iteratively approximated using a binary search. However, such an iterative approach is slow and non-differentiable. Here we show that the optimal rescaling can be found analytically using a fast and differentiable algorithm. Our algorithm works for any p-norm and can be used to train neural networks on inputs with normalized perturbations. We provide native implementations for PyTorch, TensorFlow, JAX, and NumPy based on EagerPy.

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset