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

Concept bottleneck models (CBMs) are a class of interpretable neural networkmodels that predict the target response of a given input based on itshigh-level concepts. Unlike the standard end-to-end models, CBMs enable domainexperts to intervene on the predicted concepts and rectify any mistakes at testtime, so that more accurate task predictions can be made at the end. While suchintervenability provides a powerful avenue of control, many aspects of theintervention procedure remain rather unexplored. In this work, we developvarious ways of selecting intervening concepts to improve the interventioneffectiveness and conduct an array of in-depth analyses as to how they evolveunder different circumstances. Specifically, we find that an informedintervention strategy can reduce the task error more than ten times compared tothe current baseline under the same amount of intervention counts in realisticsettings, and yet, this can vary quite significantly when taking into accountdifferent intervention granularity. We verify our findings throughcomprehensive evaluations, not only on the standard real datasets, but also onsynthetic datasets that we generate based on a set of different causal graphs.We further discover some major pitfalls of the current practices which, withouta proper addressing, raise concerns on reliability and fairness of theintervention procedure.

A Closer Look at the Intervention Procedure of Concept Bottleneck Models