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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In this paper, we present a deep neural network (DNN) training approachcalled the "DeepMimic" training method. Enormous amounts of data are availablenowadays for training usage. Yet, only a tiny portion of these data is manuallylabeled, whereas almost all of the data are unlabeled. The training approachpresented utilizes, in a most simplified manner, the unlabeled data to thefullest, in order to achieve remarkable (classification) results. Our DeepMimicmethod uses a small portion of labeled data and a large amount of unlabeleddata for the training process, as expected in a real-world scenario. Itconsists of a mentor model and a student model. Employing a mentor modeltrained on a small portion of the labeled data and then feeding it only withunlabeled data, we show how to obtain a (simplified) student model that reachesthe same accuracy and loss as the mentor model, on the same test set, withoutusing any of the original data labels in the training of the student model. Ourexperiments demonstrate that even on challenging classification tasks thestudent network architecture can be simplified significantly with a minorinfluence on the performance, i.e., we need not even know the original networkarchitecture of the mentor. In addition, the time required for training thestudent model to reach the mentor's performance level is shorter, as a resultof a simplified architecture and more available data. The proposed methodhighlights the disadvantages of regular supervised training and demonstratesthe benefits of a less traditional training approach.

DeepMimic: Mentor-Student Unlabeled Data Based Training