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

Algorithmic approaches to interpreting machine learning models haveproliferated in recent years. We carry out human subject tests that are thefirst of their kind to isolate the effect of algorithmic explanations on a keyaspect of model interpretability, simulatability, while avoiding importantconfounding experimental factors. A model is simulatable when a person canpredict its behavior on new inputs. Through two kinds of simulation testsinvolving text and tabular data, we evaluate five explanations methods: (1)LIME, (2) Anchor, (3) Decision Boundary, (4) a Prototype model, and (5) aComposite approach that combines explanations from each method. Clear evidenceof method effectiveness is found in very few cases: LIME improvessimulatability in tabular classification, and our Prototype method is effectivein counterfactual simulation tests. We also collect subjective ratings ofexplanations, but we do not find that ratings are predictive of how helpfulexplanations are. Our results provide the first reliable and comprehensiveestimates of how explanations influence simulatability across a variety ofexplanation methods and data domains. We show that (1) we need to be carefulabout the metrics we use to evaluate explanation methods, and (2) there issignificant room for improvement in current methods. All our supporting code,data, and models are publicly available at:https://github.com/peterbhase/InterpretableNLP-ACL2020

Evaluating Explainable AI: Which Algorithmic Explanations Help Users Predict Model Behavior?