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

Nowadays, the interpretation of why a machine learning (ML) model makescertain inferences is as crucial as the accuracy of such inferences. Some MLmodels like the decision tree possess inherent interpretability that can bedirectly comprehended by humans. Others like artificial neural networks (ANN),however, rely on external methods to uncover the deduction mechanism. SHapleyAdditive exPlanations (SHAP) is one of such external methods, which requires abackground dataset when interpreting ANNs. Generally, a background datasetconsists of instances randomly sampled from the training dataset. However, thesampling size and its effect on SHAP remain to be unexplored. In our empiricalstudy on the MIMIC-III dataset, we show that the two core explanations - SHAPvalues and variable rankings fluctuate when using different background datasetsacquired from random sampling, indicating that users cannot unquestioninglytrust the one-shot interpretation from SHAP. Luckily, such fluctuationdecreases with the increase of the background dataset size. Also, we notice anU-shape in the stability assessment of SHAP variable rankings, demonstratingthat SHAP is more reliable in ranking the most and least important variablescompared to moderately important ones. Overall, our results suggest that usersshould take into account how background data affects SHAP results, withimproved SHAP stability as the background sample size increases.

An empirical study of the effect of background data size on the stability of SHapley Additive exPlanations (SHAP) for deep learning models