paper-with-me

홈 › Papers

On the inductive bias of infinite-depth ResNets and the bottleneck rank

2025-01-31 · Enric Boix-Adsera

We compute the minimum-norm weights of a deep linear ResNet, and find that the inductive bias of this architecture lies between minimizing nuclear norm and rank. This implies that, with appropriate hyperparameters, deep nonlinear ResNets have an inductive bias towards minimizing bottleneck rank.

📄 PDF Abstract BibTeX arXiv:2501.19149

Code (0)

등록된 구현이 없습니다.

Tasks

Inductive Bias

Methods 이 논문이 사용한 방법론

Kaiming Initialization 설명 없음
Max Pooling Max Pooling is a pooling operation that calculates the maximum value for patches of a feature map, and uses it to create a downsampled (pooled) feature map. It is usually…
Convolution A convolution is a type of matrix operation, consisting of a kernel, a small matrix of weights, that slides over input data performing element-wise multiplication with the…
Average Pooling 설명 없음
Global Average Pooling Global Average Pooling is a pooling operation designed to replace fully connected layers in classical CNNs. The idea is to generate one feature map for each corresponding…

Similar Papers 제목 키워드 기반

The Future is Log-Gaussian: ResNets and Their Infinite-Depth-and-Width Limit at Initialization

2021-06-07 · NeurIPS 2021 12 · Mufan Bill Li, Mihai Nica, Daniel M. Roy

Theoretical results show that neural networks can be approximated by Gaussian processes in the infinite-width limit. However, for fully connected networks, it has been previously shown that for any fixed network width, $…

Gaussian Processes

Hamiltonian Mechanics of Feature Learning: Bottleneck Structure in Leaky ResNets

2024-05-27 · Arthur Jacot, Alexandre Kaiser

We study Leaky ResNets, which interpolate between ResNets ($\tilde{L}=0$) and Fully-Connected nets ($\tilde{L}\to\infty$) depending on an 'effective depth' hyper-parameter $\tilde{L}$. In the infinite depth limit, we stu…

Doubly infinite residual neural networks: a diffusion process approach

2020-07-07 · Stefano Peluchetti, Stefano Favaro

Modern neural networks (NN) featuring a large number of layers (depth) and units per layer (width) have achieved a remarkable performance across many domains. While there exists a vast literature on the interplay between…

Gaussian Processes

Neural signature kernels as infinite-width-depth-limits of controlled ResNets

2023-03-30 · Nicola Muca Cirone, Maud Lemercier, Cristopher Salvi

Motivated by the paradigm of reservoir computing, we consider randomly initialized controlled ResNets defined as Euler-discretizations of neural controlled differential equations (Neural CDEs), a unified architecture whi…

Gaussian Processes

Do ideas have shape? Idea registration as the continuous limit of artificial neural networks

2020-08-10 · Houman Owhadi

We introduce a GP generalization of ResNets (including ResNets as a particular case). We show that ResNets (and their GP generalization) converge, in the infinite depth limit, to a generalization of image registration va…

AnatomyGaussian ProcessesImage Registration