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

We present a simple but effective method to measure and mitigate model biasescaused by reliance on spurious cues. Instead of requiring costly changes toone's data or model training, our method better utilizes the data one alreadyhas by sorting them. Specifically, we rank images within their classes based onspuriosity (the degree to which common spurious cues are present), proxied viadeep neural features of an interpretable network. With spuriosity rankings, itis easy to identify minority subpopulations (i.e. low spuriosity images) andassess model bias as the gap in accuracy between high and low spuriosityimages. One can even efficiently remove a model's bias at little cost toaccuracy by finetuning its classification head on low spuriosity images,resulting in fairer treatment of samples regardless of spuriosity. Wedemonstrate our method on ImageNet, annotating $5000$ class-featuredependencies ($630$ of which we find to be spurious) and generating a datasetof $325k$ soft segmentations for these features along the way. Having computedspuriosity rankings via the identified spurious neural features, we assessbiases for $89$ diverse models and find that class-wise biases are highlycorrelated across models. Our results suggest that model bias due to spuriousfeature reliance is influenced far more by what the model is trained on thanhow it is trained.

Spuriosity Rankings: Sorting Data to Measure and Mitigate Biases