paper-with-me

홈 › Papers

Approximate Fisher Information Matrix to Characterise the Training of Deep Neural Networks

2018-10-16 · Zhibin Liao, Tom Drummond, Ian Reid, Gustavo Carneiro

In this paper, we introduce a novel methodology for characterising the performance of deep learning networks (ResNets and DenseNet) with respect to training convergence and generalisation as a function of mini-batch size and learning rate for image classification. This methodology is based on novel measurements derived from the eigenvalues of the approximate Fisher information matrix, which can be efficiently computed even for high capacity deep models. Our proposed measurements can help practitioners to monitor and control the training process (by actively tuning the mini-batch size and learning rate) to allow for good training convergence and generalisation. Furthermore, the proposed measurements also allow us to show that it is possible to optimise the training process with a new dynamic sampling training approach that continuously and automatically change the mini-batch size and learning rate during the training process. Finally, we show that the proposed dynamic sampling training approach has a faster training time and a competitive classification accuracy compared to the current state of the art.

📄 PDF Abstract BibTeX arXiv:1810.06767

Code (1)

zhibinliao89/fisher.info.mat.torch 공식 구현 pytorch

Tasks

General Classificationimage-classificationImage Classification

Similar Papers 제목 키워드 기반

An Improved Empirical Fisher Approximation for Natural Gradient Descent

2024-06-10 · Xiaodong Wu, Wenyi Yu, Chao Zhang, Philip Woodland

Approximate Natural Gradient Descent (NGD) methods are an important family of optimisers for deep learning models, which use approximate Fisher information matrices to pre-condition gradients during training. The empiric…

parameter-efficient fine-tuning

A Kronecker-factored approximate Fisher matrix for convolution layers

2016-02-03 · Roger Grosse, James Martens

Second-order optimization methods such as natural gradient descent have the potential to speed up training of neural networks by correcting for the curvature of the loss function. Unfortunately, the exact natural gradien…

Stochastic Optimization

TENGraD: Time-Efficient Natural Gradient Descent with Exact Fisher-Block Inversion

2021-06-07 · Saeed Soori, Bugra Can, Baourun Mu, Mert Gürbüzbalaban 외

This work proposes a time-efficient Natural Gradient Descent method, called TENGraD, with linear convergence guarantees. Computing the inverse of the neural network's Fisher information matrix is expensive in NGD because…

image-classificationImage Classification

Limitations of the Empirical Fisher Approximation for Natural Gradient Descent

2019-05-29 · NeurIPS 2019 12 · Frederik Kunstner, Lukas Balles, Philipp Hennig

Natural gradient descent, which preconditions a gradient descent update with the Fisher information matrix of the underlying statistical model, is a way to capture partial second-order information. Several highly visible…

Second-order methods

An iterative K-FAC algorithm for Deep Learning

2021-01-01 · Yingshi Chen

Kronecker-factored Approximate Curvature (K-FAC) method is a high efficiency second order optimizer for the deep learning. Its training time is less than SGD(or other first-order method) with same accuracy in many large-…

Deep Learning