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

AI alignment refers to models acting towards human-intended goals,preferences, or ethical principles. Given that most large-scale deep learningmodels act as black boxes and cannot be manually controlled, analyzing thesimilarity between models and humans can be a proxy measure for ensuring AIsafety. In this paper, we focus on the models' visual perception alignment withhumans, further referred to as AI-human visual alignment. Specifically, wepropose a new dataset for measuring AI-human visual alignment in terms of imageclassification, a fundamental task in machine perception. In order to evaluateAI-human visual alignment, a dataset should encompass samples with variousscenarios that may arise in the real world and have gold human perceptionlabels. Our dataset consists of three groups of samples, namely Must-Act (i.e.,Must-Classify), Must-Abstain, and Uncertain, based on the quantity and clarityof visual information in an image and further divided into eight categories.All samples have a gold human perception label; even Uncertain (severelyblurry) sample labels were obtained via crowd-sourcing. The validity of ourdataset is verified by sampling theory, statistical theories related to surveydesign, and experts in the related fields. Using our dataset, we analyze thevisual alignment and reliability of five popular visual perception models andseven abstention methods. Our code and data is available athttps://github.com/jiyounglee-0523/VisAlign.

VisAlign: Dataset for Measuring the Degree of Alignment between AI and Humans in Visual Perception