Speaker：Professor Hsing-Ming Chang (Department of Statistics, NCKU)
Topic：Improving Proximal-Newton-Type Methods for Regularized Optimization
Speaker：Professor Hsing-Ming Chang (Department of Statistics, NCKU)
Date Time：FRI. Mar 26, 2021, 10:40 AM - 11:30 AM
Place： 4F-427, Assembly Building I
Abstract
We present in this talk a method, using finite Markov chains, to obtain the run-length properties of a two-stage control process. The method furnishes the obtaining of the distribution of waiting time to signal which gives additional insight into the design and performance of a control chart when the chart is used to monitor a two-stage control process featured by a warning zone and when a departure from the null assumption can be clearly defined. For an outlined process inspection scheme assisted by a fast algorithm, numerical examples will be used to demonstrate chart performance and to verify our theoretical derivations.