Using direct kernel methods to accelerate process improvement with new and emerging technologies within and across supply chains

这项研究对于通过开发软件基础设施来帮助提高预期产出，同时减少获得更深层次理解所需的时间和成本，从而保持现有和新兴制造技术和工艺的全球竞争力至关重要。学习率的提高提供了：工艺产量的提高，性能的提高，以及设施和供应链的优化。为了实现这些目标，必须从可用数据中提取最大量的过程知识。在许多情况下，收集了大量数据，但缺乏从这些信息中提取潜在价值的适当方法。例如，在半导体代工厂制造的每个电路或MEMS器件都有大量的工艺数据。或者，随着新兴的工艺技术，如纳米技术或生物技术，很少有人了解不同的关键因素和理想的设置，以优化产品质量和产量。因此，识别和表征线性关系、非线性关系以及两个或多个变量之间的相互作用的分析平台可以加速这些过程的理解、改进、应用和商业化。

This research is critical for maintaining global competitiveness with existing and emerging manufacturing technologies and processes by developing a software infrastructure to assist in increasing desired outputs while reducing time and cost necessary to obtain a deeper level of understanding. An increased rate of learning offers: process yields improvements, performance enhancement, and facility and supply chain optimization. To obtain these goals, one must extract the greatest amount of process knowledge from available data. In many cases, huge amounts of data are collected, but suitable methods for extracting the potential value out of this information is lacking. For example, substanital process data are available for every circuit or MEMS device made in a semiconductor foundry. Alternatively with emerging process technologies such as nanotechnology or biotechnology, little is understood about the different critical factors and ideal settings to optimize product quality and output. Consequently, an analytical platform that identifies and characterizes linear relations, non-linear relations, and the interactions between two or more variables could accelerate: understanding, improvement, application, and commercialization of these processes.