This technology is developed for optimizing advanced process control decisionvirtual metrology for nano technology nodes for semiconductor manufacturing. The accuracy can reach up to 84.4, while advanced process control can be improved by 5 under the scenario of metrology delay. When new process excursion occurred especially when ramping up new technologiesproducts, the monitoring rules can be redefined for intelligent fault detectionclassification via CNN transfer learning to effectively shorten the self-learning cycle time. It can balance the productivityproduct yield while effectively shortening yield rampingmass production time.

"This novel technology has the following breakthroughs:
 
 1. The empirical study has shown that advanced process control can be improved by 5 when the prediction accuracy can reach up to 84.4.
 
 2. Since it does not rely too much on domain knowledge, it can reduce the timecost of model parameter adjustment. It can be widely used in different process fields to balance productivityyield effectively while shortening the time for yield ramping.
 
 3. Redefining the monitoring rules for intelligent fault detectionclassification via CNN transfer learning to effectively shorten the self-learning cycle time when new process excursion occurred especially when ramping up new technologiesproducts."

This technology aims to optimize the performance of advanced process control via analyzing manufacturing big data by multi-mode process sensors to develop decision-based virtual metrology technology. This novel technology can be effective deployed to various manufacturing steps by integrating the confidence level of VM predictiondomain knowledge to derive appropriate decisions for real-time control. Manufacturing industries limited to item-by-item measurement restrictions can employ this technology to balance productivityyield effectively while shortening the time for yield ramping.