paper-with-me

홈 › Papers

Towards understanding the true loss surface of deep neural networks using random matrix theory and iterative spectral methods

2020-01-01 · ICLR 2020 1 · Diego Granziol, Timur Garipov, Dmitry Vetrov, Stefan Zohren, Stephen Roberts, Andrew Gordon Wilson

The geometric properties of loss surfaces, such as the local flatness of a solution, are associated with generalization in deep learning. The Hessian is often used to understand these geometric properties. We investigate the differences between the eigenvalues of the neural network Hessian evaluated over the empirical dataset, the Empirical Hessian, and the eigenvalues of the Hessian under the data generating distribution, which we term the True Hessian. Under mild assumptions, we use random matrix theory to show that the True Hessian has eigenvalues of smaller absolute value than the Empirical Hessian. We support these results for different SGD schedules on both a 110-Layer ResNet and VGG-16. To perform these experiments we propose a framework for spectral visualization, based on GPU accelerated stochastic Lanczos quadrature. This approach is an order of magnitude faster than state-of-the-art methods for spectral visualization, and can be generically used to investigate the spectral properties of matrices in deep learning.

📄 PDF Abstract BibTeX

Code (0)

등록된 구현이 없습니다.

Tasks

GPU

Methods 이 논문이 사용한 방법론

Average Pooling 설명 없음
ReLU How Do I Communicate to Expedia? How Do I Communicate to Expedia? – Call ☎️ +1-(888) 829 (0881) or +1-805-330-4056 or +1-805-330-4056 for Live Support & Special Travel…
1x1 Convolution A 1 x 1 Convolution is a convolution with some special properties in that it can be used for dimensionality reduction,…
Batch Normalization 설명 없음
Bottleneck Residual Block A Bottleneck Residual Block is a variant of the residual block that utilises 1x1 convolutions to create a bottleneck. The…
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…
Residual Block Residual Blocks are skip-connection blocks that learn residual functions with reference to the layer inputs, instead of learning unreferenced functions. They were introduced…
Kaiming Initialization 설명 없음

Similar Papers 제목 키워드 기반

Appearance of Random Matrix Theory in Deep Learning

2021-02-12 · Nicholas P Baskerville, Diego Granziol, Jonathan P Keating

We investigate the local spectral statistics of the loss surface Hessians of artificial neural networks, where we discover excellent agreement with Gaussian Orthogonal Ensemble statistics across several network architect…

Deep Learning

Random matrix theory and the loss surfaces of neural networks

2023-06-03 · Nicholas P Baskerville

Neural network models are one of the most successful approaches to machine learning, enjoying an enormous amount of development and research over recent years and finding concrete real-world applications in almost any co…

Geometry of Neural Network Loss Surfaces via Random Matrix Theory

2017-08-01 · ICML 2017 8 · Jeffrey Pennington, Yasaman Bahri

Understanding the geometry of neural network loss surfaces is important for the development of improved optimization algorithms and for building a theoretical understanding of why deep learning works. In this paper,…

Memorization

Universal characteristics of deep neural network loss surfaces from random matrix theory

2022-05-17 · Nicholas P Baskerville, Jonathan P Keating, Francesco Mezzadri, Joseph Najnudel 외

This paper considers several aspects of random matrix universality in deep neural networks. Motivated by recent experimental work, we use universal properties of random matrices related to local statistics to derive prac…

A Random Matrix Theory Approach to Damping in Deep Learning

2020-11-15 · Diego Granziol, Nicholas Baskerville

We conjecture that the inherent difference in generalisation between adaptive and non-adaptive gradient methods in deep learning stems from the increased estimation noise in the flattest directions of the true loss surfa…

Deep Learning