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

Locally interpretable model agnostic explanations (LIME) method is one of themost popular methods used to explain black-box models at a per example level.Although many variants have been proposed, few provide a simple way to producehigh fidelity explanations that are also stable and intuitive. In this work, weprovide a novel perspective by proposing a model agnostic local explanationmethod inspired by the invariant risk minimization (IRM) principle --originally proposed for (global) out-of-distribution generalization -- toprovide such high fidelity explanations that are also stable and unidirectionalacross nearby examples. Our method is based on a game theoretic formulationwhere we theoretically show that our approach has a strong tendency toeliminate features where the gradient of the black-box function abruptlychanges sign in the locality of the example we want to explain, while in othercases it is more careful and will choose a more conservative (feature)attribution, a behavior which can be highly desirable for recourse.Empirically, we show on tabular, image and text data that the quality of ourexplanations with neighborhoods formed using random perturbations are muchbetter than LIME and in some cases even comparable to other methods that userealistic neighbors sampled from the data manifold. This is desirable giventhat learning a manifold to either create realistic neighbors or to projectexplanations is typically expensive or may even be impossible. Moreover, ouralgorithm is simple and efficient to train, and can ascertain stable inputfeatures for local decisions of a black-box without access to side informationsuch as a (partial) causal graph as has been seen in some recent works.

Locally Invariant Explanations: Towards Stable and Unidirectional Explanations through Local Invariant Learning