论局部群体风险证书 / On Local Population-Risk Certificates
1️⃣ 一句话总结
本文提出了一种局部风险证书方法,通过构建当前模型周围候选更新的置信区间,帮助我们判断模型更新是否会降低风险,从而在保证风险不增加的前提下决定是否接受更新。
This paper develops local certificates for population-risk increments around a current model. For a local candidate set \(\mathcal D\), the certificate is a two-sided confidence band for \(P({\ell_{\theta+v}-\ell_\theta})\) over \(v\in\mathcal D\). As an application, the upper endpoint of this band yields a risk-controlled update rule: an update is accepted only when its certified upper endpoint is nonpositive; otherwise the current model is retained.
论局部群体风险证书 / On Local Population-Risk Certificates
本文提出了一种局部风险证书方法,通过构建当前模型周围候选更新的置信区间,帮助我们判断模型更新是否会降低风险,从而在保证风险不增加的前提下决定是否接受更新。
源自 arXiv: 2606.19147