paper-with-me

Papers

Explicit Regularisation in Gaussian Noise Injections

2020-07-14 · NeurIPS 2020 12 · Alexander Camuto, Matthew Willetts, Umut Şimşekli, Stephen Roberts, Chris Holmes

We study the regularisation induced in neural networks by Gaussian noise injections (GNIs). Though such injections have been extensively studied when applied to data, there have been few studies on understanding the regularising effect they induce when applied to network activations. Here we derive the explicit regulariser of GNIs, obtained by marginalising out the injected noise, and show that it penalises functions with high-frequency components in the Fourier domain; particularly in layers closer to a neural network's output. We show analytically and empirically that such regularisation produces calibrated classifiers with large classification margins.

📄 PDF Abstract BibTeX arXiv:2007.07368

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Asymmetric Heavy Tails and Implicit Bias in Gaussian Noise Injections

2021-02-13 · Alexander Camuto, Xiaoyu Wang, Lingjiong Zhu, Chris Holmes 외

Gaussian noise injections (GNIs) are a family of simple and widely-used regularisation methods for training neural networks, where one injects additive or multiplicative Gaussian noise to the network activations at every…

Differential Privacy of Gaussian Process Posterior Sampling

2026-06-16 · Tomasz Maciazek arxiv

We study the privacy of releasing posterior sample paths from a Gaussian process (GP) when the entire training set including covariates and responses is private. Unlike standard differential-privacy (DP) mechanisms that …

On the Inherent Regularization Effects of Noise Injection During Training

2021-02-15 · Oussama Dhifallah, Yue M. Lu

Randomly perturbing networks during the training process is a commonly used approach to improving generalization performance. In this paper, we present a theoretical study of one particular way of random perturbation, wh…

Patch-based learning of adaptive Total Variation parameter maps for blind image denoising

2025-03-20 · Claudio Fantasia, Luca Calatroni, Xavier Descombes, Rim Rekik

We consider a patch-based learning approach defined in terms of neural networks to estimate spatially adaptive regularisation parameter maps for image denoising with weighted Total Variation (TV) and test it to situation…

Binary ClassificationDenoisingImage DenoisingModel Selection+1

Enhanced Denoising and Convergent Regularisation Using Tweedie Scaling

2025-03-07 · Naïl Khelifa, Ferdia Sherry, Carola-Bibiane Schönlieb

The inherent ill-posed nature of image reconstruction problems, due to limitations in the physical acquisition process, is typically addressed by introducing a regularisation term that incorporates prior knowledge about …

DenoisingImage Reconstruction