Okay. So in this particular weeks, we talked about regression and we talked about support-vector machines. So today, many people have learned these two subjects, either in statistics course, in machine learning course or some other related course. It's also interesting that many people learn these two subjects without learning OR in advance, okay? So that's definitely possible, even if you have no idea about that this square optimization thing, even if you have no idea about convex function, even if you have no idea about Katie conditions. Well, you may still understand the existence of these two models, and you may still invoke some modules to solve these problems, all right? Because today, modules, libraries, they are all somewhat be available everywhere. So it's not so difficult to write a few lines of codes to import some libraries and invoke them and solve the problems that you want to solve, okay? Linear regression all kinds of libraries, SVM a lot of libraries. You don't really need to know the underlying theory before you may use the models, that's fine. So that's why many people don't really understand the underlying optimization theory behind these two models, but they still use them every day. I won't say it's wrong, because if that can help you solve, your problem is fine. It's just that if you are the person that you want to improve the efficiency or the accuracy for using these models, sometimes you need to know the underlying theories. Or if you move somewhat more advanced, if you want to improve these models, if you want to invent better models, then you really need to know those theories, right? So we teach you theory not because they may be helpful for you to use these models, no. To use these models, to do applications what you need to know are the application things. You don't really need to know the theory. To know the theory in my belief is because we want to find better models, you want to construct designed, developed better models. So that we need to know how this works and then we may know, we may decide how to improve the way it works, okay? So whenever we have a course teaching about theory, we don't want to just let you know how it works. We want eventually let you know how to improve the existing thing. So that may not be easy, but some people must do that, okay? If everybody just want to do applications, if everybody just want to use some tools that other people invent, we have no more new tools to be used. So there must be some courses talking about theories, we don't need to have a lot, of course, talking about theories. We don't need to have a lot of people working on theories, but we need some people to do that. So somehow, if you someone like me, if you enjoy doing these mathematics, if you enjoy doing these models talking about the theories, looking at the algorithms, try to do analysis and try to find a better way to do things. Maybe you are also the appropriate person to devote yourself to the development of new theory, to the development of new tools that may be very useful for others. So it may not be easy, you need a lot of mathematical foundations, algorithm foundations, whatever foundations. But once you are able to do a little bit thing, the impact would be huge, because all future students in OR, they need to look at your findings. So that is not impossible, think about George Dancer, think about about William Crush, that's always possible. If I cannot do it, maybe you can do it, if you cannot do it, maybe your students can do it, it's always possible. So hopefully you like theory and hopefully you like the materials. Hopefully, in the future, some of you may also devote yourself to the development of theories. If that's the case, it would be very nice. [MUSIC]