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

Understanding the interpretation of machine learning (ML) models has been ofparamount importance when making decisions with societal impacts such astransport control, financial activities, and medical diagnosis. While currentmodel interpretation methodologies focus on using locally linear functions toapproximate the models or creating self-explanatory models that giveexplanations to each input instance, they do not focus on model interpretationat the subpopulation level, which is the understanding of model interpretationsacross different subset aggregations in a dataset. To address the challenges ofproviding explanations of an ML model across the whole dataset, we proposeSUBPLEX, a visual analytics system to help users understand black-box modelexplanations with subpopulation visual analysis. SUBPLEX is designed through aniterative design process with machine learning researchers to address threeusage scenarios of real-life machine learning tasks: model debugging, featureselection, and bias detection. The system applies novel subpopulation analysison ML model explanations and interactive visualization to explore theexplanations on a dataset with different levels of granularity. Based on thesystem, we conduct user evaluation to assess how understanding theinterpretation at a subpopulation level influences the sense-making process ofinterpreting ML models from a user's perspective. Our results suggest that byproviding model explanations for different groups of data, SUBPLEX encouragesusers to generate more ingenious ideas to enrich the interpretations. It alsohelps users to acquire a tight integration between programming workflow andvisual analytics workflow. Last but not least, we summarize the considerationsobserved in applying visualization to machine learning interpretations.

SUBPLEX: Towards a Better Understanding of Black Box Model Explanations at the Subpopulation Level