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

We propose a generic framework to calibrate accuracy and confidence of aprediction in deep neural networks through stochastic inferences. We interpretstochastic regularization using a Bayesian model, and analyze the relationbetween predictive uncertainty of networks and variance of the predictionscores obtained by stochastic inferences for a single example. Our empiricalstudy shows that the accuracy and the score of a prediction are highlycorrelated with the variance of multiple stochastic inferences given bystochastic depth or dropout. Motivated by this observation, we design a novelvariance-weighted confidence-integrated loss function that is composed of twocross-entropy loss terms with respect to ground-truth and uniform distribution,which are balanced by variance of stochastic prediction scores. The proposedloss function enables us to learn deep neural networks that predict confidencecalibrated scores using a single inference. Our algorithm presents outstandingconfidence calibration performance and improves classification accuracy whencombined with two popular stochastic regularization techniques---stochasticdepth and dropout---in multiple models and datasets; it alleviatesoverconfidence issue in deep neural networks significantly by training networksto achieve prediction accuracy proportional to confidence of prediction.

Learning for Single-Shot Confidence Calibration in Deep Neural Networks through Stochastic Inferences