Canonical Likelihoods: A New Likelihood Treatment of Nuisance Parameters
提出一种基于似然函数奇异值分解的新方法，用于处理含多余参数的模型，并展示了其在重参数化下的不变性。
A new approach to the likelihood analysis of models with nuisance parameters is proposed, based on a singular value decomposition of the likelihood function, and is applied to several two-parameter models. The invariance of the approach under reparameterization is demonstrated, and the paper concludes with comments on multiparameter models.