Evaluating Deep Taylor Decomposition for Reliability Assessment in the Wild

Publikation: Bidrag til tidsskriftKonferenceartikelfagfællebedømt

Dokumenter

We argue that we need to evaluate model interpretability methods 'in the wild', i.e., in situations where professionals make critical decisions, and models can potentially assist them. We present an in-the-wild evaluation of token attribution based on Deep Taylor Decomposition, with professional journalists performing reliability assessments. We find that using this method in conjunction with RoBERTa-Large, fine-tuned on the Gossip Corpus, led to faster and better human decision-making, as well as a more critical attitude toward news sources among the journalists. We present a comparison of human and model rationales, as well as a qualitative analysis of the journalists' experiences with machine-in-the-loop decision making.
OriginalsprogEngelsk
TidsskriftProceedings of the International AAAI Conference on Web and Social Media
Vol/bind16
Sider (fra-til)1368-1372
ISSN2162-3449
DOI
StatusUdgivet - 2022
Begivenhed16th International AAAI Conference on Web and Social Media - Atlanta, USA
Varighed: 6 jun. 20229 jun. 2022

Konference

Konference16th International AAAI Conference on Web and Social Media
LandUSA
ByAtlanta
Periode06/06/202209/06/2022

ID: 339852192