 Oh, I made a mistake. All right. That's the second one. Yeah, take your time. Don't worry. Okay, let's make it perfect. You just need to screen full screen. No, it is the second part. This is part two. Yes. Yes. Oh, this is part two. Yes. Yes. You need a part one. Yes. I made a mistake to put along. Is this okay? It's coming. Coming. Yeah. Yes. We have the outline now. Part one and part two. Perfect. Yes. Hello, everybody. My name is Sang Muk Lee. I am speaking, coming to you from South Korea Seoul. It's a little past 4am here. I'm going to talk about inverse methods in the era of machine learning and deep learning. I'm going to tell you from my educational point of view, you know, for those who want to do both geophysics and the new latest deep learning and stuff. I would like to thank the organizers and in particularly Alec very well, very much because he has made this talk possible for me. He has been gracious to invite me. I give a lot of inspirational talks around the world. So these days since last, since coronavirus, I did not travel very much. So my English has got a little bit rusty. So I cannot speak as fast as before. So this coronavirus has come to my tongue. This is the outline of my talk first. I'll talk a bit about myself because like Alec mentioned in 2006, I had a devastating accident. I became quadriplegic. That means I cannot move anything below my neck, not a single finger. So, but luckily, I mean, life has unexpected twists and turns. And this made me a better person and not that I was a bad person, but made me realize things to gain a bigger perspective of life. So in a way you can call it an awakening. Okay. And believe it or not, I became very famous in Korea. In Korea, they call me Stephen Hawking of Korea. All the media and everybody calls start calling me Stephen Hawking of Korea so much so that when an English friend visit me in Korea, I have to tell him that remind him that in Korea you have to call him Stephen Hawking of England, so that people will not get confused. This is a joke. Anyway, so because of my accident and my return back to work. In around 2010, I was able to start a big national program in Korea. And it was worth about 10 million US dollar a year. And I realized that it's really difficult to spend money when you get such a big grant. So I set up a computational science program at our university. The only way that professors can spend money is by setting up programs and departments. As you know, computational science at that time was all about numerical analysis, finite element, finite difference, all those numerical simulation and stuff. But nowadays, it's more about data science. Now, I used to be a, I am still a field scientist, but because of my injury, I had to reinvent myself into computational science and then I had to deal with data science, numerical analysis and then all these things buzzword that's popping up. So I tell you how I got around this complication. First, I didn't know what object oriented programming was, but it was rather simple than I thought. And in the final part, I would like to mention to you about some of the things I got money for my money was not for computational science but to raise the level of. To raise the level of Korean, the quality of life for people with disability in Korea. So one of the things that I really recently found out works is that voice coding. I cannot use a single finger, but I can still do coding using voice. It's more simple than regular natural language. And as you know, these days, an important buzzword keyword is Jedi Jedi stands for justice equality diversity and inclusion, and in Korea I became the champion of Jedi. And so we, I'm also helping to train many people with disability with severe disability like myself, not just normal disability, really hard severe disability to get job and find career to see will be the first part. The second part is about our science. It's why do you science is to broaden your knowledge and science is all about observation due data and things. In 2015, I was able to change Korea in a big way up to that point for about 25 years Korean academic research community did not have access to very big high tech state of the art research vessels. This was possible in Europe, United States, Japan, but not in Korea. Everything was monopolized by government. But after I became very famous in 2015, I was called to testify against the government at National Assembly, you know, our parliament. And there, I beat up the government so much that, well, that a new law was passed that now allows all academic community all the research fleet in Korea to be shared with the university. And this was a big turn, big contribution. And so I will. Yeah, I mean I'm telling you this because you have to know this to understand what I'm why I am emphasizing observation and data. And then this part to physical invert theory in the era of machine learning and deep learning. Well, I'm a fan of deep learning machine learning, but at the same time, a bit skeptic. The reason is that we never have enough data data. Machine learning deep learning works well when there is enough data, you know, if there is enough data, different types of algorithm actually do not matter that much it's the amount of data but when you are in the real science where you're trying to discover something you were never given enough data. So, so this is my biggest skepticism and my biggest shortcoming. I'm blessed. I mean, the reason I'm not really blessing deep learning machine learning. The reason is that, as you know, you know, these things can work very well on the image. So, in earth science we have satellite observation atmospheric science, where they all gather image satellite image and stuff. It works well on the image, but in more diverse applications. Sure. I'm not fully convinced. So there are. So my message is that I'm waiting for this amalgamation of geophysical invert theory and data science machine learning to become more fruitful in the future. So, you know, my luck has been continuing. And last year, I received a big award from our university. You know, every university, the, the chancellor of the, or the president, their job is to improve the ranking of the university while they're in tenure. Our university at the moment is about 37th in the world. I mean, every every university has different systems. So we don't really care whether we are 3750 only thing we care is the gap between Korea and Japan. So I realized that Japan, the University of Tokyo is 22nd. So we only have to catch up with them. That's the only concern anyway. So they selected 10 departments, you know, and are funding them to make this 10 departments within global rank of 10. Well, I'm not sure whether I'll be able to do this, but I got the funding. And when I got the funding. And one of the big corporation in Korea wanted to do something for environment, and they, you know, supported in big way the funding so I'm this age able to expand the vision so that I care about writing paper with my students but in a way inspiring young people in Korea and around the world to become part of this big event to understand our planet Earth. This planet, this project is named Planet A. Do you know why? Well, I got the name because one of the UN Secretary General Ban Ki-moon during a conference said global greenhouse gas reduction conference said, Well, when it comes to global gas reduction, greenhouse gas reduction, there is no plan B because there is no planet B. So I got the name. So this planet A, which is funded by the university and also by the big corporation is giving me, it's something that I have to deal with. And to do so, I'm trying to organize data hackathon and also develop instruments. I tell you, instruments are very important new way to measure Earth. Okay. Okay. Before and after the accident, I was like, I like before my accident, enjoying cruises going to Antarctic spending like three months on average of three months at sea in remote places around the globe. But then, in 2015, I had an accident, which crippled me to the filter. But thanks to Stephen Hawking of England, publics, Korean public said, Oh, that's Stephen Hawking. So I, with this, with his, with the, with his thanks to his effort, I became a national icon immediately. So I have many things to thank to ICTP because I heard that Stephen Hawking used to visit there quite frequently. Well, I returned to Korea about in 1990, 1998, and I was used to fundamental basic blue sky type research because I was studying mid ocean ridge, which is something out in the deep sea nowhere. Unless you have, there is no economic benefit. You know, you're studying it like, like other planet. And so, at that time, Korea was developing country and I was appalled because Korea didn't care about the basic science and was also interested in economic mineral resources. So, come on, you know, this is a stupid way to spend taxpayers money, I thought, but I was alone. But after 20 years, I could see Korea changing, you know, Korea boasts as the only country in the world that was a developing country that received foreign aid and is now able to become a donor country where we are a member of global developed country where we contribute annually. We are the only country in the world. So, with that meant that Korea changed from developing to develop country status, we're on the margin, but anyway, and I realized that there are two things that we should not confuse. One is policy, science policy. Science policy is dictated by economists, by government budget planners and all this hiring people. So, science policy, scientists, it's like a national agenda. It's very hard to put our opinion. What we need is policy for science. This is different. It sounds same, but I think it's completely different. I gave a talk at AUGS recently telling people that don't expect too much from national scientific policy, but we have to work together to make policy for science and scientists. That's what we want. Well, this morning I ran another webinar online webinar because 20 years ago, 30 years ago, I complained that Korea was not spending much money on basic research like the ocean ridge. But now, since last year, I'm the chair of global inter-age community. So, you see how things have changed, and I know many students here, but you can follow our activity in www.inter-age.org. Now I said I spent 12 years abroad in Europe and US, and then seven years at a government lab doing global ocean research operations. And then I returned to my university, my Anna Mara, in 2003. And this is when I returned back to Korea, teaching. Real active moving geology in Korea was very difficult. Korea is a very old geology. Unlike Japan or Taiwan, which is rattled, we used to be like that 200 million years ago. This tectonic front has moved now to Japan and Taiwan. So, yeah. So, in a way, Korea is a very boring geology. Also, the country is very heavily vegetated, so I could not inspire students to study geology, earth science, geophysics, and et cetera. What do I do? So, with a professor that I knew at Caltech, we went to California. California, everything is basically a desert, and California is recent geology. So, I took Korean students together with geological field trip team of Caltech, and we did wonderful, very inspiring geological field trip. And this was taken a week before my accident. And this is what happened a week later, and I'm really, I'm still stuck underneath this driver's seat, and somebody had the nerve to take the picture, which I'm very grateful to explain what has happened to me. The roof of the car van collapsed, and so I was the only one that was injured because I stood up straight because of I was holding the handle. So, the accident that I suffered is called SCI, not scientific citation index, but spinal cord injury. As a result of accident, I had a disconnection, so they had to do a operation to put back my neck and body together. And so it was called neck fusion operation, and so as a result of this accident, I became paralyzed neck down. Well, it's the symptom is different from Stephen Hawking, he is, he has Lou Gehrig, I'm different, I'm spinal cord injury. And I can speak, he cannot, but still the public imagination just, I just fed into the public imagination. So as a result, I cannot move anything below my neck. So in terms of these days, computer science wise, the injury, the problem that I have is a network connection. So the CPU and the remaining part is not connected. And as I tell as a joke to people, if you want, if you don't want this to happen to you, drive Volvo. Okay, but this accident could have really devastated my life. But it was an awakening moment for me. Actually, I felt very calm and my life completely changed. I had no denial depression or anything. So but, but when you have this near this accident, you start to question about big things in life. What could I have missed if I have died that day on that side. Well, I could have resorted to religion. But as a scientist, we are trying to solve problems and address problems. So this scientific trade got into me and I said, I want to know I don't want to believe. And I think this word was made iterated by the famous scientist car saying and he said, I don't want to believe I want to know. So, although we will never find this answer. I said, if I stop questioning this. It's like science neglect for scientists, like, because the water is deep. I'm not going to research and study this. It's like that. So this has been my obsession. And this is where the accident occurred. And I had a chance to revisit this site. Six years later. And I laid a flower for a undergraduate student who died on that day. Now as you. I cannot flip a page of a book, you know, I cannot use my hand, but because I can control the computer with this red device, which is called integral mouse sip and puff device this small device. I can manipulate computer and on computer I can look at PDF listen to Kindle books. I mean look at the Kindle books read Kindle books audio books. So, I realize I did not lose very much, you know, I don't miss miss very much because now I have all the knowledge that I want right in front of me, you know. And then I said, Wow, isn't like amazing. I'm really fortunate and lucky to be injured at the right time in human history, when there is science that can support me. Oh, I never had a gloomy day. Oh, I'm asking my caregiver to give me a water because one of the injury that outcome is that I don't get saliva pumping of saliva back to my mouth is really has hampered a lot. So I started reading a lot of books I tried to find answers so not just in geophysics and other science but from all history, religion, literature, everything. I just wanted to find out what other people as said about life and what I can learn from them. So one of the biggest hurdle obstacle in my life as a scientist is distraction on other subject. So, because it's so easy to Google, click up and other subjects I just forget about the research paper that I have to write and follow that trail, you know, this is, I mean that means that in this newly connected world. Knowledge is, if you know where to look, it's, and if you have the will that's everywhere. In addition to just reading books I did a lot of traveling on one case with television crew. I took 40 days across the United States. And this was televised in Korea, and it made me even more famous. The right bottom corner is when I, my final destination was from Los Angeles to New York and when I arrived at United Nations in New York, all the presses around the world gathered and start making, taking photos and my writing my story. And, well, on average, I make 10 overseas trips a year since the last 12, 13 years. So I'm, you know, so this is the first time in 15 years that I am staying more than six months in Korea. You know, one of the memorable place that I visited was Merida, Mexico in 2013, there was a IAGA meeting in Merida. The Merida was chosen because it was an impact site of that big meteorite that killed all the dinosaurs, you know. Well, actually if you go here and that site is covered with sediment so you cannot see but with gravity, because there is a difference in density between sediment and hard rock behind below. So you can see the shape of the impact crater that changed history. So, you know, I like IUGG because they go around the world and have very nice meetings at this interesting sites. Again, what happens if I go to countries like Mexico, I gave this guy the money to build a wooden ramp like this, and I was able to get on and off. And I visited this very famous place and the reason was that I was curious how a civilization that sacrifices its own people to maintain, I mean, can keep on going for a certain period. And I found out that this is a very barren area so when there is a war, they don't, they just enslave the normal people, farmers and because they need them for production, but to reduce the mouth they have to feed, they have to sacrifice all the noble people. Well, they say we cannot donate this gift, having this tainted blood, we have to give the blood of the noble and pure guys so in that such. So, I realize there was a, I mean that explains many things. I was able to link up with one of the novelty, no royalty in United Arab Emirates, her in Sharjah, Princess Jamila Al-Kashmi and together with the funding from LG, we were able to run for three years, something called education, EDUCOM, which stands for assistive technology. These are technology that is needed for people with disability, EDUCOM stands for education and competition, and we had a very good competition running with Korean students and those at United Arab Emirates. Here you have the Korean team and I found out this was so successful to my wildest dream. The reason was that these students are a fan of Korean pop culture and they want to come to Korea and this, because the winners are provided opportunity to visit Korea, we were very, very successful to even our wildest imagination. Usually we select one team, but one year there was so much infighting that we had to select three teams. And two years ago, a small city in Korea named Pyongchang hosted the Winter Olympics. It's the smallest city to ever host the Olympics and the governor of that state asked me if I would make this small city like Davos in Switzerland, meaning that I would make a big event focused on earth science and global issues to attract people and one of the things that I proposed was that, you know, the theme of this work was, I mean, science is about breaking unchartered territory. So, as you can see, it was called at the limit of our knowledge, staring into the future. This was the theme that I got and this has stopped because there was so much, because after success, so much infighting against the government officials that they start infighting that it ended up abruptly after only two years. Anyway, in 2012, I mean, when I tell people I have near-death experience and people ask me, how was it like? How was it like? And in 2015 or 2012 or so, I had the chance to visit Paris, IPGP. This is where Alik used to work before, a very famous geophysical institution, IPGP. And because there was a sudden strike, I had to cancel some trip and then I had several days to kill. And I read a book written by Sarah Bakewell, who tells the story about Montaigne. Montaigne is the guy who created a literary form called SA and he had a near-death experience and this changed his life. And before he was involved in politics, local politics, the public, but then he, after this experience, he confined himself to writing about his daily events, his thoughts. I mean, before Montaigne, people wrote something, they had a purpose, but SA, if you think about it, is a literary form where you don't have an object. It's like Facebook, modern-day Facebook. So I wanted to really feel connected because instead of asking me how it felt like, please read this book because it's really amazing that across 450 years of time, I share, I know what he means, and he would know what I felt. So it's like time machine. And so I said I had to go to Bordeaux and I took a train and went to Bordeaux, stayed there three days, and then found out there was a statue of Montaigne. I took the picture just before the train. I went to Bordeaux to feel the breath of Montaigne. Of course, I enjoyed the wine, but also that was not the purpose. As I mentioned earlier, in 2010, I received almost $10 million a year project for four years. And I created, I mean, I am by nature a sea-going field geologist, but I created a computational science program in undergraduate, and also with the money that I had, I basically merged the graduate program. So now, in the last 10 years, I'm in charge of computational science program at Seoul National University. This kind of program should be resided within mathematics department as part of applying mathematics. But because of my ability to raise funds and my public recognition, I'm still running this program. So this is taking a lot of my time running both graduate undergraduate program and computational science. So this is, I just show you some of the basic curriculums. The red here is a prerequisite. And so I teach a course called understanding of computational science and I teach Python, C++, and then I teach a very interesting course called actually a history of history and foundation of computational sciences. This is to give an opportunity to people even in social department or humanity who are afraid of mathematics. And by showing them how this thing was made, I'm trying to overcome prejudice, like it's called dummy for computational science or something like that. And I really enjoyed teaching this course because teaching also allowed me to gain greater and deeper perspective on mathematics from number theory all the way to modern day mathematics. You know, actually, at first the course was named history of mathematics, then many people's professors in mathematics department complain, how, how, why, why does professor in earth science teach our area. So I changed the name to history and foundation of computational science, but it's basically mathematics until Newton, cows and oiler. After that, I think there's a split in computational science before it's basically same as normal mathematics. So it. Oh, I basically take the book of John Steelwell. And I see a breakpoint where computational science become its own field. And this was more than 10 years ago and this was before the rush of the event of data science, as we know today. So, many people were interested in Fortran C++ because, and I used to learn from Fortran, and many people ridicule me, but you know, you still need C in Fortran, because it matters for some application that we need speed. Python and all this data science are wonderful, but they're not necessarily the best when it comes to speed. And they're very easy low level entry, but for us, you know, serious scientists. For us, the speed of computation matters and it's used in many fields, including subduction, a mental convection, plate tectonics and all this stuff. Of course, another area of high speed computing is visualization. And now comes data science. This is a late comer, but I teach these things. And I wonder how useful, how does it really help us? Well, and I also looked into I was attracted by complex phenomena. So I spent some time with my student looking at something called agent based modeling which can, which defines problem into interaction between agents. And you can define the but I quickly ran into trouble because modeling you can do with computer, but how do you know how do you measure all this parameter that comes from complex system. This was I hit a wall. Okay. So this understanding and better understanding of complex phenomena in nature and society is still a big challenge. In our computation science, we can produce whatever result we want, but the problem is that it has to fit the observation. How do you guarantee that's the prediction. That's the future. To do that you need much, much more observation and there is a reason why I keep the model simple, you know, so this was I learned the lesson. Okay. Now, I am, I started, I started programming around early 80s and start only doing it seriously when I went to graduate school in US in around 1986. And people tell me how do I keep up with this all this knowledge, you have to know to teach. And so I tell you one example of how I reinvented myself. Well, this is somewhat related with history. I'm an old school. So I've seen how the computer has developed and become a magic, I mean, more mysterious person but I think everything is. For instance, when I start programming in Fortran, I did not understand what object oriented was. It was very difficult to get my head around because I knew too much about other methods, you know, I had too much bias. Well, then I one day I realized it's, it's a very simple matter. It's about stack and heap pointer. You understand what stack and heap is and then the pointer, everything, the complexity and diversity of modern computing is nothing, you know, have you seen such calculator. When I first came to us, I found out all the students were using this HP 15 calculator. What do you mean? Perhaps no more but so the way it works is unlike Cassio or standard calculation, it use stack, which is how the computer operates. So when you want to add a time multiply three at 10 and 15 and at 10 to 15 and multiply by three. This is what it does. We usually said 10 plus 15 times and then get the result and time three but how the stack works is that you put three. You put 10, you put five and you add and then the bottom to the first two is edit and then you multiply then you get. Okay. Stack is like pancake to get to the bottom, you have to pass through all this layers on top. Now this became a problem. When you write a problem, you define variables and you stack it. And, for instance, there's a big chunk of data like array or big messy thing. Then it's like piling up too much pancake. So what people did was that, oh, it came with an idea pointer reference. Let's just put the, put everything in the heap where there's enough memory. I mean, heap and stack is not physically just logical definition of memory but let's put it somewhere. And let's just have a small address pointing to the first part. When you start going through the pancakes, it, you don't have to flip through all those array index one by one. Okay. Then people got smarter is it all. Let's put heterogeneous data there. And so, once they did that, they said, okay, we can put, we can have a pancake that represent the start of the, that thing object and do everything in a normal way. So this small idea of pointer became larger and larger. And now, you know, when you write a program, everything that's big, like class object or heterogeneous data structure and everything is now. And the pointer and people went even further crazy and then decided to make various data type custom made data type. So once I understood this, hey, it's not very different from the standard for trend that I learned. Nothing has basically changed. Many things have changed, but I could understand that. There is a clear difference between enterprise computing and scientific computing and Google and all these big companies their enterprise computing their job is making money being successful. I can compare it to liking pure mathematics in pure mathematics, you can solve any, any problem. But for us in applied mathematics, we have to solve issues that are relevant to scientific community and society. We have to do scientific computing to apply mathematics in a sense that we have to stay on the track. I cannot just do something because it makes money. So, so I want to go tell you in the second part about the inverse theory. The inverse theory as I mentioned earlier requires a lot of data. And once you have enough data, like those generated by cell phone and our movement. Algorithm basically is more or less become same. Well, in science, we never have enough data. My experience shows, and as you all know, works very all on image telling the difference between cat and dog and object detection on images, but in earth science, there are fields like satellite image and certain areas where image processing is important. No doubt that machine learning deep learning works well, but does it apply to other areas where the situation is more complicated. I don't think so at the moment, you know, because as I will tell you later, there are some problems. Excuse me, this is Alec. My question, would you like to have a break or break will be in a few minutes. Yes, I can have a break. I can have a break now. Now, okay, wonderful. Yeah, it would be good. Yeah. Okay, very good. Then we will have a break but not very long. Please. At 10 o'clock, we will reconvene. Now it is a 955 Central European time indeed. And we will reconvene in five minutes, please. Thank you very much. It would be enough. Yes, for five minutes break. Okay.