Unverified50% confidenceFactExact time
Equal Opportunity Difference衡量不同群体在正样本上的真正例率差异，Demographic Parity衡量不同群体获得正向决策的概率差异
1
Sources
50%
Confidence
Long-term
Relevance
7/12/2026
First Seen
Sources
AIML Capstone Project Selection: Is a Multi-Agent Debate System Worth Building?
redditr/learnmachinelearning7/11/2026
Related Claims
Unverified配对设计通过在同一个体上观测两种处理的差值将个体差异从误差项中剥离，等价于将组间方差转化为组内比较以提高统计功效76% similarUnverifiedOn the issue of gender bias alignment, different fairness standards such as "equality of opportunity" and "equality of outcomes" may point to mutually contradictory model behaviors.68% similarUnverifiedThe IER pays attention to disparate impact when evaluating discrimination—even if a company has no explicit discriminatory policy, a systematic disadvantage to a particular group reflected in hiring outcomes may trigger an investigation.61% similarUnverified位置偏差是偏好数据集中一个被广泛研究的系统性问题，人类标注者评估并排回答时会受展示顺序影响59% similarUnverified学术研究表明预测市场在估算事件概率方面往往优于单一专家预测，因为它聚合了大量异质信息58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/491561API
curl https://kongchang.com/api/v1/knowledge/claims/491561MCP
get_claim(id=491561)