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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

A convolutional neural network (ConvNet) is usually trained and then testedusing images drawn from the same distribution. To generalize a ConvNet tovarious tasks often requires a complete training dataset that consists ofimages drawn from different tasks. In most scenarios, it is nearly impossibleto collect every possible representative dataset as a priori. The new data mayonly become available after the ConvNet is deployed in clinical practice.ConvNet, however, may generate artifacts on out-of-distribution testingsamples. In this study, we present Targeted Gradient Descent (TGD), a novelfine-tuning method that can extend a pre-trained network to a new task withoutrevisiting data from the previous task while preserving the knowledge acquiredfrom previous training. To a further extent, the proposed method also enablesonline learning of patient-specific data. The method is built on the idea ofreusing a pre-trained ConvNet's redundant kernels to learn new knowledge. Wecompare the performance of TGD to several commonly used training approaches onthe task of Positron emission tomography (PET) image denoising. Results fromclinical images show that TGD generated results on par withtraining-from-scratch while significantly reducing data preparation and networktraining time. More importantly, it enables online learning on the testingstudy to enhance the network's generalization capability in real-worldapplications.

Targeted Gradient Descent: A Novel Method for Convolutional Neural Networks Fine-tuning and Online-learning