After learning the theoretical foundations of CyberGIS and Geospatial Data Science with a particular focus on spatial computational domain for estimating computational intensity and we have a couple of applications that illustrate the utility of such theoretical foundations, and you could imagine based on the diversity of Geospatial Analytics. You see the potential power of such theoretical foundations for helping us figure out optimal solutions to large-scale Geospatial Analytics that could be increasingly data and computation intensive. For future trends, we want to bring back four knowledge domains we covered at the beginning of the course. These four knowledge domains include computation and data intensive applications and the sciences including what we have just to learn the theoretical foundations allow us to solve computation data intensive problems using divide and conquer strategies and there's theoretical guidance from computational intensity point of view. Geospatial Data Science has the potential for richer theories to be developed, taking into account spatial characteristics. The CyberGIS represents both science and technology that need to be innovated further to enable the applications and the sciences that are Computation and Data Intensive, guided by the theories, concepts, methods, prompt Geospatial Data Science. Of course, the fourth knowledge domain is Advanced Computing and the Cyberinfrastructure that will likely continue to change rapidly. The application driven nature is quite clear. We have lot of problems coming from the top and technologies are also moving very fast going forward. It's enabled by the rapid advancement of Advanced Computing cyberinfrastructure and a [inaudible] on top of that CyberGIS. We want to cover one by one across this four knowledge domains using some examples, and also provide some broader contexts for us to be able to see what might be coming in the future. For a long time, in the history of science, we have three key modalities of scientific work. Experiment, observation, and theory. This has been practiced for hundreds of years and until more recently, Computation, roughly over half-century has become a major new way of approaching scientific research and enabling scientific frontiers. In the past few decades, big data, data science have really become another new modality of scientific research. This is sometimes termed as data-driven discovery and innovation and only in the past decade, artificial intelligence, machine intelligence have researched to be effective methodology and technology for scientific research. Now, if you look at this three new modalities of scientific research: AI, Big Data, and Computation, this really happened very quickly. Now, we have this loop we need to close for a lot of exciting Geospatial applications and science problems. To close this loop is the biggest challenge and opportunity we're facing in the context of Geospatial Discovery and Innovation. The CyberGIS and Geospatial Data Science have made their roles to play going forward. What are the big scientific and a societal challenges we are facing? Well, to name a few examples here: Disaster & Emergency, Food Safety & Security, Health & Wellness, Environmental & Water Sustainability, Earth & Energy Resources. In all these domains of challenges and problems, we need Geospatial approaches. In fact, Geospatial Data & Software are very much situated in the center of this convergence among analysis, modeling, and the sensing approaches. Sometimes, we perform Analytics, sometimes, we have models, often times simulations. Sometimes we go out to take measurements, have tons of sensors, but all of these approaches are increasingly dependent on geospatial software tools and dependent on geospatial data science to make progress, to inform the crosscutting problems in this donut diagram on the blue color part, covering the five examples I just mentioned. We've got to address increasing geospatial problems and questions in all of these big scientific and societal challenges. That's how exciting this is, and going forward it's only going to be more exciting from a computation, data intensive point of view, but also to solve these big challenges, we need collaboration capabilities. That's one of the major elements of CyberGIS evolution, is to support collaborative problem-solving beyond individual level problem-solving. What would be the new scientific modalities embedded in collaborations? We need to come up with new capabilities for CyberGIS, is a major direction going forward for both scientific research as well as for potential innovation opportunities for CyberGIS. With respect to geospatial data science, one of the four knowledge domains that is situated under competition and data intensive applications in the sciences. We understand the intersection among three broad knowledge areas that is informing the development of geospatial data science. This includes the three areas such as geospatial sciences and technologies, mathematical and statistical sciences, cyberinfrastructure, and computational sciences. But in the intersection, the combination of CyberGIS, artificial intelligence, and data science form the core of geospatial data science through the learning of the concept of conditional intensity, by constructing spatial competition domain. We covered two examples, one has to do with the distribution of points. For instance, the density of the points plays a key role in impacting the computational intensity, and in the viewshed analysis example, there is in fact, spatial dependence because certain parts of the trail will block your views if you be situated somewhere on a piece of landscape. Your views are blocked because where you are situated has to do with how you would be looking at the surrounding areas. The dependence of where you are with regard to what you want to see also is important to be taken to account for estimating your computational intensity. In fact, the terrain characteristics such as the peak pit and the flat, we learned, have a role to play in impacting your computational intensity for the viewshed analysis. You see the spatial characteristics have important roles in shaping your spatial competition domain for understanding the computational intensity and to also guide the divide and the conquer strategies once you know well about such special characteristics. There are other examples of spatial characteristics, such as interaction, heterogeneity, and so on. But such spatial characteristics are important to evaluate for the benefits of computational performance. In this case, for the theoretical foundations of computational intensity, we compared computational complexity versus intensity. These are two different concepts. In particular, computational intensity is a concept that is very much rooted in the consideration of spatial characteristics versus computational complexity has little to do with spatial characteristics in many of the algorithmic problems. From a computational point of view, there are concepts and considerations such as uncertainty versus validity, performance versus reliability. You want to, for instance, achieve your analytics to be faster, but at the same time, you want to assure your results are going to be reliable. This is a typical trade-off you need to consider in your computational context. A major opportunity and challenge for geospatial data science going forward is this scale. It cuts across both in the spatial realm as well as in the computational realm. We're going to tackle much bigger scale problems from the spatial side, as well as on the computational side. How do we connect the dots between the spatial and the computational characteristics and the considerations? Increasingly, we also need to be concerned about multiscale representations. Again, both in the spatial side as well as in the computational side. There's a lot unknowns as we go forward with our aspirations for solving bigger scale problems, both spatially and computationally. That's where new concepts, theories, and methodologies need to be developed for competition, as well as a geospatial data science at a scale. Now, the knowledge domain of Advanced Computing and the cyberinfrastructure is quite exciting because of the innovation pace is so fast, technologies are getting developed very rapidly, so if you would be asked, how many zeros are in a quintillion? Specification and you get an answer like this. There will be 18 zeros. Now, I need to tell you, the fastest computer in the next a few years could perform a quintillion calculations per second, you might be amazed. That's how fast a computer could go soon, and you might say, "oh, how could I do with this fastest computer with my own geo-spatial problems? Well, roughly 10 years ago, your smartphone was the fastest computer. You're current smartphone could be the fastest computer 10 years ago. In about 10 years from today, a computer that will conduct quintillion calculations will be with you possibly all the time. How do we innovate CyberGIS and Geospatial Data Science to solve a variety of problems and enable numerous applications? Anticipating this computing power will be with us individually everywhere. It is such an exciting exercise to do for both the research and technical communities. One example is there's also potential breakthrough for quantum computing to be widely adopted. In fact, there are quantum computers developed in both research and industrial settings. We know the current architecture of computing is one human computing architecture, that is based on the input-output central processing unit and all this conventional architecture that links the different parts of your computers working together based on this zero and one binary specification internal to your computer chips. Quantum computing is a different because it's in a sense, probabilistic computing. The internal state of quantum computing chip is based on zero and one, and everything between the zero and one is probabilistically occurring that covers a very large space of representation of your computing gates from the hardware point of view. Apparently, this is an advantage for quantum computing to sift it through many, many possible options very quickly. This is a new, very exciting computing paradigm, different from our Von Neumann computer architecture based on fixed zero and one state to represent the possibilities of the options you need to compute through. So there's earlier work on the spatial optimization problem solving using quantum computing. There's interesting promise from this piece of work that indicates if quantum computing could be widely adopted and we would see tremendous potential for solving some really difficult spatial optimization problems and possibly some other geospatial analytics problems that could benefit from quantum computing. Again, computing frontiers continue to evolve very rapidly. The conventional computing based on harnessing parallelism, that's the current frontier of high-performance computing, and possibly new frontier based on quantum computing, could both benefit geo-spatial discovery and innovation through the development of CyberGIS and geo-spatial data science. Finally, what's going to be the future of CyberGIS? With the rapid innovation of advanced computing and cyberinfrastructure, with theoretical guidance from geospatial data science, with many exciting problems and challenges from variety of application and the scientific domains. CyberGIS remind you of the definition, geospatial information sciences and systems based on advanced computing and cyberinfrastructure. It has just about a decade of history. This is just a very small sample of the papers I'm involved in the past decade showing trajectory of this field. It's still a very young field, but it's very exciting to see much bigger potential of CyberGIS, as we are contextualizing from the bottom up that technology enablement from advanced computing and cyberinfrastructure, but also many scientific and application drivers from the top, motivating further innovation and development of CyberGIS. Certainly, there's a rich and fertile field of geospatial data science, particularly providing the opportunities for coming up with new theories and methodologies in a spatially explicit fashion. Overall, I think CyberGIS is going to evolve into ecosystem which has a number of areas that are interconnected to each other, motivated by a variety of applications and science research cases, and enabled by advanced cyberinfrastructure capabilities and services. In this course, we have covered some very useful technical aspects of CyberGIS and geospatial data Science. You have learned geovisualization, you have learned geospatial data processing, you have learned MapReduce programming model for geospatial big data analytics, and we've covered the theoretical foundations for dividing and conquering large data sets to inform optimal geospatial analytics in a computationally scalable fashion. Now in this broader ecosystem, we know there are good number of parts we need to go into depth, for instance, the interactive and real-time analytics part is very exciting because we now have lot of geospatial data streams. They are coming very fast. Traditional analytics are not able to directly support such data streams as this data streams could become quickly outdated by the new data streams coming to be more relevant than the data streams already in the past. So how do we come up with the optimal as well as appropriate right analytics to sift through such data streams and make most out of the data available, is a very interesting and exciting topic. To going into a topic like this, we would need another course, likely more advanced to going into the technical depth to understand how do we employ the most cutting edge advances of CyberGIS and geospatial data science to understand what would be the opportunities and the possibilities of such real-time and interactive analytics that could be useful in many different applications. Again, for the other areas, we would likely need to have much more in-depth treatment into those areas. Collectively, they are forming a broader CyberGIS ecosystem. That's how exciting this could be, and thank you for taking this course. It's been a great pleasure of interacting with you. There's huge potential in the future, but for getting in touch with us both to ask questions about this course as well as for future opportunities, I have listed the contact information on this slide for both me and my colleague, Dr. Anand Padmanabhan. E-mail is the best way to get a hold of us offline, and we look forward to our further interactions. Thank you again.