MACHINE LEARNING BASED LOGANALYSIS FOR AUTOMATED ANOMALY DETECTION
Many sensors are used in a single production process, making it difficult to pinpoint the exact source of a problem. More than one process cycle is required to make a semiconductor wafer. There are many cycles in this process, and it is difficult to spot abnormalities in time, thus the process continues until it is complete. The cost of producing these wafers is high, and a process failure can have a significant impact on both time and money. As a result, anomaly detection in semiconductor production can benefit greatly from machine learning. A manufacturing facility may interrupt the operation and fix the problematic equipment if irregularities in the production process could be discovered or predicted sooner. As a result, semiconductor producers would see an improvement in process yield and a reduction in expenses.
machine learning, based log-analysis, automated, anomaly detection