English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

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

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