AI Paper Atlas

Research map / Learning theory and algorithms / Fairness

Bridging Fairness and Explainability: Can Input-Based Explanations Promote Fairness in Hate Speech Detection?

Yifan Wang, Mayank Jobanputra, Ji-Ung Lee, Soyoung Oh, Isabel Valera, Vera Demberg

ICLR 2026 · Learning theory and algorithms › Fairness · approach: Causal and policy fairness in decisions

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use L2-aggregated or Occlusion explanations to detect biased predictionsBERT and Qwen3-4B · Civil Comments / Jigsaw bias detectionfairness correlationImproved
aggregate embedding attributions by mean instead of L2 norm (ablation)explanation-based bias detection · Civil Comments bias detectionfairness correlationHurt
use validation-set sensitive-token reliance to select the fairest modelmodel selection · Civil Comments, 7 encoder and 4 decoder modelsSpearman rank correlation / MRR@1Hurt
add sensitive-token reliance regularizer during trainingBERT · Civil Comments race/gender/religionFNR disparity and other fairness metricsImproved
Grad L1 explanation regularization for gender biasBERT · Civil Comments genderaccuracy and fairnessImproved
tune alpha on a fairness-balanced metric instead of task accuracy (ablation)explanation-based debiasing · Civil CommentsfairnessImproved
combine traditional and explanation-based debiasingBERT · Civil CommentsfairnessImproved
apply traditional debiasing methods (group balance, CDA, dropout, etc.)BERT, Qwen3-4B · Civil Commentsfairness metrics: BERT race Avg_iu 3.17 -> 0.50 (CDA), FNR disp 10.04 -> 18.45Mixed
explanation-debias with attention-based methods then detect bias with attention explanationsexplanation-debiased BERT · Civil Comments racefairness correlationHurt
detect bias with Occlusion and L2 explanations in explanation-debiased modelsexplanation-debiased BERT · Civil Comments racefairness correlationNo clear effect
replace explanation-based detection with LLM-as-a-judge (ablation)Qwen3-4B predictions · Civil Comments racefairness correlation: 0.194 -> 0.104 (Qwen3-4B, K=5)Hurt
use larger GPT-OSS-120B as judge instead of Qwen3-4B (LLM decision) (ablation)LLM-as-a-judge · Civil Comments raceAvg_iu biased/unbiased: 0.065/2.59 -> 4.42/2.35Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy