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

Deep neural networks (DNNs) often rely on massive labelled data for training,which is inaccessible in many applications. Data augmentation (DA) tackles datascarcity by creating new labelled data from available ones. Different DAmethods have different mechanisms and therefore using their generated labelleddata for DNN training may help improving DNN's generalisation to differentdegrees. Combining multiple DA methods, namely multi-DA, for DNN training,provides a way to boost generalisation. Among existing multi-DA based DNNtraining methods, those relying on knowledge distillation (KD) have receivedgreat attention. They leverage knowledge transfer to utilise the labelled datasets created by multiple DA methods instead of directly combining them fortraining DNNs. However, existing KD-based methods can only utilise certaintypes of DA methods, incapable of utilising the advantages of arbitrary DAmethods. We propose a general multi-DA based DNN training framework capable touse arbitrary DA methods. To train a DNN, our framework replicates a certainportion in the latter part of the DNN into multiple copies, leading to multipleDNNs with shared blocks in their former parts and independent blocks in theirlatter parts. Each of these DNNs is associated with a unique DA and a newlydevised loss that allows comprehensively learning from the data generated byall DA methods and the outputs from all DNNs in an online and adaptive way. Theoverall loss, i.e., the sum of each DNN's loss, is used for training the DNN.Eventually, one of the DNNs with the best validation performance is chosen forinference. We implement the proposed framework by using three distinct DAmethods and apply it for training representative DNNs. Experiments on thepopular benchmarks of image classification demonstrate the superiority of ourmethod to several existing single-DA and multi-DA based training methods.

A General Multiple Data Augmentation Based Framework for Training Deep Neural Networks