U.S.-France Cooperative Research: Construction and Evaluation of Accuracy Estimators

这个为期三年的奖项支持康奈尔大学的乔治·卡塞拉和巴黎第六大学理论与应用统计实验室的克里斯蒂安·罗伯特在统计和概率方面的美法合作研究。这项研究涉及从实验中得出的统计推断。为了解决这个问题，他们建议进一步发展精度估计的程序。数据后统计推断是基于实验中的所有信息，包括对数据的了解。建议的研究将整合和评估使用决策理论和条件推理的数据后推理方法。他们将研究估计器在不同损失函数下的行为和损失函数鲁棒性的基本问题，并探索将决策理论与随机模拟技术相结合的问题。合作研究项目利用了互补的兴趣和专业知识。这位美国研究者将他在决策理论方面的专业知识带到这次合作中。这是由法国调查员在计算和算法统计方面的丰富经验补充。它们的共同努力可能产生新的统计计算技术和方法，包括更好的事后数据推断程序和计算精度措施的有效方法。

This three-year award supports U.S.-France cooperative research in statistics and probability between George Casella of Cornell University and Christian Robert of the Laboratory for Theoretical and Applied Statistics, University of Paris VI. The research is concerned with statistical inferences from experiments. In addressing this problem they proposed to further develop procedures for accuracy estimation. Post-data statistical inferences are based on all information in an experiment, including the knowledge of the data. The proposed research will integrate and evaluate approaches to post-data inferences using decision theory and conditional inferences. They will investigate fundamental questions on the behavior of estimators under different loss functions and in loss function robustness and explore problems of assimilating decision theory with stochastic simulation techniques. The cooperative research project takes advantage of complementary interests and expertise. The U.S. investigator brings to this collaboration his expertise in decision theory. This is complemented by the French investigator's extensive experience in computational and algorithmic statistics. Together, their efforts may result in new techniques and methodology of statistical computing, including, better procedures for post data inferences and efficient methods for computation of accuracy measures.