An International Workshop on Bayesian Data Analysis

贝叶斯数据分析国际研讨会PI：David Draper Proposal：DMS 0311766摘要过去10年，随着基于模拟的计算方法的广泛采用，贝叶斯统计方法在能力、科学相关性和适用性方面都取得了长足的进步，其中马尔可夫链蒙特卡罗技术是进步的引擎。尖端的贝叶斯建模现在正在广泛的领域进行，包括生物信息学、计算机科学、工程学、流行病学、机器学习和统计学。跨学科的思想交流很重要，但越来越难实现，因为每个领域的文献如雨后春笋般涌现，对类似概念使用不同的术语，阻碍了搜索引擎帮助一个领域的工作者跟上其他学科的最新信息的能力。贝叶斯方法本身对研究很重要，但对于最大的科学相关性(1)对建模的仔细关注和(2)方法与背景相适应的需要也是至关重要的。在这个项目中，加州大学圣克鲁斯分校(UCSC)新成立的应用数学和统计系(AMS)内的统计组将通过主办2003年8月8日至10日星期五至周日在加州大学圣克鲁斯分校(UCSC)举行的贝叶斯数据分析国际研讨会来解决这些问题，作为一种卫星会议，将于2003年8月3日至7日在加利福尼亚州旧金山举行的联合统计会议(JSM)之后立即举行。我们的目标是聚集来自世界各地对贝叶斯建模和数据分析感兴趣的大约100人，其中包括大约30名受邀参与者(领先的贝叶斯研究人员)作为演讲者和讨论者，以促进对共同问题的多学科讨论和跨学科思想转移(其中许多人将是年轻的研究人员)。讲习班的重点将是贝叶斯数据分析：从科学或决策中的一个实际问题开始，用统计学的术语表述这个问题，使用贝叶斯方法解决原始问题，并从统计和实质性两个方面讨论解决方案的优点和缺点，同时充分注意现实世界背景和贝叶斯模型建立、检查和重新制定之间的相互作用。NSF的资金将主要用于为年轻研究人员、少数族裔和其他代表性不足的群体提供部分旅行和当地费用支持。

An International Workshop on Bayesian Data AnalysisPI: David DraperProposal: DMS 0311766 AbstractBayesian statistical methods have gained considerably in power, scientific relevance, and applicability over the last 10 years, with the widespread adoption of simulation-based computation methods involving Markov chain Monte Carlo techniques as the engine of progress. Cutting-edge Bayesian modeling is now going on in a wide variety of fields, including bioinformatics, computer science, engineering, epidemiology, machine learning, and statistics. Interdisciplinary transfer of ideas is important but increasingly difficult to achieve, as the literatures in each field burgeon and the use of different terms for similar concepts hampers the ability of search engines to help workers in one field keep up-to-date in other disciplines. Bayesian methodology is important to study in its own right, but for greatest scientific relevance (1) careful attention to modeling and (2) the need for methodology to be appropriate to context are also crucial.In this project the Statistics Group within the newly-forming Department of Applied Mathematics and Statistics (AMS) at the University of California, Santa Cruz (UCSC), will address these issues by hosting an International Workshop on Bayesian Data Analysis at UCSC from Friday through Sunday 8-10 August 2003, as a kind of satellite meeting to be held right after the Joint Statistical Meetings (JSM) nearby in San Francisco, CA, from August 3--7, 2003. Our objectives are to bring together about 100 people interested in Bayesian modeling and data analysis from a wide variety of disciplines around the world, including about 30 invited participants (leading Bayesian researchers) as speakers and discussants, to promote multidisciplinary discussion of common problems and interdisciplinary transfer of ideas (many of these people will be young researchers). The focus of the Workshop will be Bayesian data analysis: starting with a real problem in science or decision-making, formulating this problem in statistical terms, using Bayesian methods to solve the original problem, and discussing the strengths and weaknesses of the solution both statistically and substantively, with plenty of attention to the interplay between the real-world context and the Bayesian model building, checking, and reformulating. NSF funding will be principally used to provide partial travel and local expenses support for young researchers, minorities, and other under-represented groups.