 our JTF capable headquarters in all of the AFRICOM area of responsibility. We now have full administrative control and training readiness authority over the 173rd Airborne Brigade and we ensure that the 173rd is trained and ready for crisis response and contingency operations to include the ability to support the North and West African response force that we call the NARF. And finally, we're responsible for the oversight of all Army activities and personnel in Italy who is a critical NATO ally. So that's a big mission with a big footprint and we rely on the work of our teammates in the intelligence enterprise to make sense of this environment and to make better decisions faster than our competitors and our adversaries. Today you will hear from one of our operational subordinate commands, the 207th Theater Military Intelligence Brigade. The 207th conducts military intelligence analysis, collection and exploitation in support of CTAP, VAP and AFRICOM. It's one of 17 brigades belonging to US Army Intelligence and Security Command, INSCOM. And we're really proud to have a close relationship with INSCOM and we leverage the intelligence capabilities that the 207th brings to the fight to accomplish our mission in Africa. What's remarkable is that the 207th is the smallest theater MI Brigade in all of INSCOM but it has the widest coverage. A full spectrum of intelligence support that connects and delivers the intelligence enterprise to the AFRICOM Theater. That's an area about three and a half times as big as the continental United States with 53 sovereign nations in five regions, each with a unique history, culture and political system. It's a challenging environment for Intel collection. It's an area where our near pair of competitors are actively seeking to influence events on the ground through diplomacy, disinformation, economic leverage and military presence, much of it malign. An area where violent extremist organizations continue to be a persistent threat, destabilizing fragile governments and expanding their safe havens while opening space for our competitors. The 207th helps CETAF, AFRICOM, understand these activities and the range of threats to US interests and those of our African partners. Today you will hear about one of our unique capabilities. The Africa Data Science Center or ADSC is a pilot program that leverages data science to help analysts quickly understand and exploit the massive amount of data that is currently available to the intelligence community. This capability is a pilot as a game changer for us. The ADSC enables traditional intelligence collection methods in Africa's resource constrained environment where we can't always rely on ISR assets. This program supports one of the Secretary of the Army's primary objectives to ensure that the Army becomes more data-centric and can conduct operations in contested environments. What is really exciting is this approach is exportable to the rest of the Army in the DOD enterprise. And in particular to other areas of responsibility where the US military anticipates and in fact is operating in contested environments every day. We see a lot of opportunity for this program and I know that INSCOM is looking at opportunities to expand its reach. I now want to turn the floor over to the real expert, Colonel Mark Denton. He'll provide a more detailed view on how the ADSC program actually works and then we'll have an opportunity for a few questions and answers and we look forward to that dialogue. Thanks very much for your attention and please welcome Colonel Denton. Thank you, sir. Good afternoon, ladies and gentlemen. As General Wassman pointed out, I'm Colonel Mark Denton, commander of the 207th Military Intelligence Brigade. And today I'll be talking to you about the Africa Data Science Center. But before we get started, one of the individuals I also have here with me is Mr. JT Gill. JT is my senior civilian. And he's been around since we started with the Africa Data Science Center. Let me get this tablet up and then we will continue. Well, it's a good thing that I have a backup because the tablet is down right now. So JT, we'll go to the notes. So as General Wassman said, I am Colonel Mark Denton and I am the commander of the 207th Military Intelligence Brigade. Today we're going to be talking about the Africa Data Science Center. And as I talk about that, one of the things I do want to let you know is that we are headquartered in Vicenza, Italy as the 207th Military Intelligence Brigade. Not a bad assignment there in Vicenza, Italy. And one of the things I'll also make sure that you understand is it is late in the day here, but it's even later in Italy. So jet lag is kicking in right now. So bear with me this evening and we'll get started. So again, as General Wassman mentioned, we are one of 17 of INSCOM, the Intelligence Security Command's brigades that's out there. And we are the Theater Intelligence Brigade that is assigned to the Africa Command Area of Responsibility and further operationally command by the Southern European Task Force, as you just heard from General Wassman. Our Warrior Corner, again, is focused on using data science as an intelligence multiplier. Our Africa Data Science Center, or the ADSC, as we will call it, is not on our EMTO, Modified Table of Organization and Equipment. So that is an out-of-hide activity that we've had to do. And it's currently a three-person team within our analysis and control element, our ACE, which is an element within our operations battalion. Go to the next slide, please. It's the situation. So as you can see on this slide, it shows the chart of our ADSC. The ADSC, again, is led by a captain and a warrant officer and supported by three contract data science engineers. But before we dive into that, what I do want to do is I do want to play this video real quick for you. So go ahead real quick and roll the video, please. And turn it up as live as you can. What's the situation? Historic first. And in response to ruffles, unprovoked, it was a federal government, a dissent, and it was our nation, a US nation, a government, a city women of the United States, a federal government, a state, a state, a state. But it's a non-competitive task force from Africa. But it was done by an ADSC. Hey, I can help you narrow this down. We are the Africa Data Science Center, transforming data into decision advantage. The Army G2's relevant deep sensing at the edge warriors corner. You heard Alex Miller as he artfully and eloquently articulated that we don't have a data problem is what he said, we have, you know, what do we do with the data problem, which is much different. Our intelligence analyst has a data overload problem, as you saw there inside the video. So therefore, the Africa Data Science Center has been important to the workflow and production for our analysts, as you heard General Wassman also artfully articulate earlier. A little bit of background on the Africa Data Science Center. The ADSC is an INSCOM pilot program which started in 2018 and continues today with INSCOM, AFRICOM, and CTAF funding that important capability. So General Wassman mentioned a little bit, but let me just tell you Africa lies at a global crossroads. It holds tremendous geo-strategic significance while being shaped by competing forces of prosperity, poverty, and also conflict that we see on a daily basis. Across the continent, all 54 countries, CTAF and the 207th MI Brigade Enterprise provide cutting edge information that regularly impacts national level decisions on a daily basis. The 207th MI Brigade Africa Data Science Center uses machine learning techniques to quickly scrutinize large sums of data, eliminate redundancies, and provide relevant information in a usable format. The ADSC leverages best practices from across the intelligence community and applies leading edge data science techniques to fulfill operational intelligence requirements to operational commanders like General Wassman within our theater. As an out of high data science support cell, as previously mentioned, with a startup mentality, this unique program received recognition by the House Armed Services Committee in the FY22 National Defense Authorization Act. The NDAA lauded the ADSC for serving as a model of innovation and providing invaluable clarity on adversary activities across Africa. As General Wassman mentioned, this capability is not specific just to Intel or specific just to CTAF. This capability can be exported to other environments and other theaters as well. Yesterday, in the opening ceremony, we heard the Secretary of the Army lay out her priorities. The ADSC is one small but innovative example of how the 207th is meeting the Secretary of the Army's objective number two to become more data-centric at echelon. We accomplished this by using machine learning, advanced analytics, data visualization tools, and most importantly, intelligence domain expertise to build products that enable analysts to sort through vast amounts of data, spot trends easier, and ask more informed questions to get to the so-what side of the equation a lot quicker. As you saw in the video, our analysts can be overwhelmed by the volume and variety of data available to them on a daily basis. One way the ADSC has tackled this problem is by building software categories and curate tens of thousands of reports for analysts in just a few minutes. This would normally take analysts days or weeks to organize that amount of information before analyzing it. As an example, the ADSC can easily consolidate over 3,000 reports in a matter of seconds. Go to the next slide, please. As we talk about transforming data into the intelligence process, the ADSC is not a stand-alone. The ADSC is embedded and incorporated into the intelligence cycle and everything that we do on a daily basis as it integrates with the analysts and not separated from the analysts. So ultimately, the ADSC is focused on using modern tools to solve complex intelligence process. To do this, we had to adapt the traditional software development practices to fit the Army intelligence cycle, as you see there depicted on the slide. By integrating data science into the intelligence process at various echelons of command, the ADSC supports both analysis and operations through tight integration with the CTAF commander's requirements or whatever your entity is, their requirements and the objectives within our analysis and control element in our battle rhythm events. So this is scalable and taleable to your specific requirements and environment and to your commander's needs. For instance, the ADSC helps senior leaders in the Africa community of interest better understand complex problems such as the evolving violent extremist organization threat and near peer adversary Russia and China and their activities within the Africa area of responsibility. One of the most critical parts of integrating advanced analytics with intelligence is understanding the problem. Without deep domain expertise, i.e., the analysts, we're going to develop things that end up off the mark. So the analysts have to be involved in what you develop as a data science or data scientist. The analysts have the expertise, while the data scientists have programming expertise. So they're not one and the same. So you do need that expertise on both sides. Before we start using machine learning, coding, and big data techniques, we have to understand our customer's requirements or our commander's requirements and needs. Consideration for handling sensitive data permeate every step of data science integration from how we get the data to what happens after we transform, disseminate, and store that data. While many of the Africa data science tools are used on TS systems or JWICS, the information often used that is often used at the collateral level and can be made appropriate at any other level, again, based on the needs and requirements of the commander. The ADSC uses existing platforms where feasible and we're not feasible. They also can create their own versions of a platform. The Africa data science is an enabler first and foremost. Our team integrates our work with the IC. So along the intelligence community-wide platforms and partners, we will utilize those first. So whether it's FADEMIS, Bode, Ejema, Safehouse, Getz, ArcGIS, Chrome, some of the more familiar crate, et cetera, some of those more familiar tools and platforms, that's what we tend to use on a routine and daily basis to be able to integrate with the rest of the IC. The ADSC pushes and pulls data in various formats while transforming it en route to meet the needs of the production requirement of our consumer. Our soldiers are not equipped or trained to do that task. So the Intel professional is not trained at Fort Wachuka to be able to do the work of the engineers or the scientists. So I don't want you to confuse that we're now turning our soldiers, our 35 foxes, that's trained at Fort Wachuka into data scientists and we can talk about that a little bit more. In addition, although the African Data Science Center isn't a traditional int, again it's not an int by itself and we can have that debate about open source also, but we take all that information that's available out there and it's not a traditional int by itself. However, and it doesn't produce and disseminate traditional intelligence products, but they do build tools to support and enable the analysts in ways you can't buy off the shelf. The ADSC does not replace the analysts again, rather they work with the analysts who contribute deep regional expertise which cannot be replaced by a scientist. Go to the next slide please. Here's a vignette and again as you heard in the video we talk about going from chaos to clarity on the right hand side there you see all that purple stuff, that's all the stuff that comes in that an analyst needs to, that's all the chaos and an analyst needs to make sense of that so we can inform folks like General Wassman. So another major tenet of the Africa Data Science Center is work to simplify and organize the tremendous amount of data available to our organization. The idea is to get our analysts to the starting line a lot quicker, allowing them to focus on analytical part of intelligence work and let computers do the manual repetitive task. So we'll do that through machine learning. So time and time again we heard from analysts that it can be difficult to accurately assess trends with so much data available out there to them on a daily basis. You saw the analysts again in the video with all that information coming in and he's got it, folks just asking him what's the answer, what's the answer? Well he can't begin to start doing the analytical work because he's just sifting through data and oftentimes a lot of that data is the same repetitive reports that are coming in but they're coming in different forms from different entities. The graphic on the slide shows the chaos that too much data can bring. The ADSC uses a series of machine learning models to automatically characterize thousands of reports for analysts, excuse me, allowing them to focus on analytical task at hand and get out of the blocks a lot faster. This machine learning system uses natural language processing which is essentially pulling in elementary terms, turning words into numbers and then doing the math immediately after that to put intelligence reports into categories based on mathematical similarities to a baseline. That's the easiest way to explain that. We then take those curated reports and let the analysts interact with the data in a way that makes sense to them, either through a dashboard or an Excel spreadsheet, whatever format they prefer. This goes back to where I started with the comments about sense, making sense and then acting. So you sense the data, you try to make sense of it by binning it or categorizing it and then you're able to give it to commanders to be able to act on that data based on the predictive analysis that comes from that. This program was developed at CTAPF by the ADSC with the direction from analysts and the theater leadership and is designed to help our folks at the big picture of what's happening on the continent within the context of army doctrine and also our intelligence trade calf. So trying to nest all those things together and this can be done in any theater, whether you're in the Pacific or whether you're in UCOM or other theaters, you can do the same exact thing. So as I get on short final here and try to wrap up, let me just recap what I really just said. If you took nothing else away in the last 10 minutes or so, what I'm telling you is in the last 10 minutes, the U.S. AFRICOM AOR is a vast and complicated environment that possesses significant challenges to U.S. interests and requires unique processing methods to gain the information advantage. When we talk about spread across such a large array of environment over 54 countries, sometimes 55 depending on who you ask on what day, it is challenging to be able to get all that information in a usable form to our commanders on a daily basis. And the Africa Data Science Center is a force multiplier in CTAPF's arsenal to exploit and maximize tangible insights from the massive amount of data currently available to the intelligence community. Emerging technologies continue to increase the amount of data the joint force has at its disposal making previously inaccessible insights a daily occurrence. The rapidly developing field of data science will provide decision advantage to commanders at the unified command and theater echelons. Theater embedded teams ensure technical expertise is co-located with regional subject matter experts, allowing for quick prototyping of unique data science solutions to meet the commander's evolving information requirements. As I said, in closing, the Africa Data Science Center has a bright future. With the increased intelligence production, it allows we've only begun to scratch the surface of the ADSC's capabilities. In the future, we would like to see data sharing through technical reach back to a hub and tie in with other ADSCs across the enterprise in more of a hub and spoke model. Right now, we are essentially a silo and there aren't other ADSCs out there where we can tie into a hub and spoke model. But right now, I'll tell you though, the INSCOM G7 under the direction of INSCOM is working through that. And I know some folks at Fort Wachuk are also working this where they're exploring concepts for an enterprise approach to Africa Data Science Center and more of a data science hub across the entire community. So the capability to process so much data independently through machine learning is the way ahead that will allow analysts to spot emerging trends and support commander's intelligence requirements much quicker. In the lessons learned since its inception in 2018, the Africa Data Science Center program can be applied to other national security and regional challenges. So let me just say one final comment and then we'll open up for questions. And I'll also turn it over to my senior civilian see if he's got any comments he wants to make. But I do know that folks out of Fort Wachuka, General Hale is making a lot of progress on this in terms of making our individuals more data literate. And that's really important. So the leaders out there, the lieutenants, captains, even the soldiers as they go through training, they're getting a lot of this information on understanding data and how to exploit and utilize that data and becoming more data literate. So there are a lot of folks that's making a lot of efforts on this and thank you for your time. I really appreciate it. JT, do you have anything before we close? Go to the next slide, please. Nothing additional at this time, sir. Okay. We'll go ahead and open up for any questions that anyone has. Can we get a microphone? Hi, Todd, South Army Times. I have two questions real quick. One, you mentioned that this is exportable. You also mentioned it was siloed because it's not a hub necessarily. So I guess first off, do these exist in other combat commands yet they're just not connected or is this solely an enterprise only for Africa and not replicated elsewhere at all? So for us right now, and I'll have JT also comment because JT's been doing this. He's been part of it since 2019. But for us, we are not connected to other data science center. We're connected to other individuals like folks in the soft community. They have a similar capability, but there isn't a hub at the top that has the ability like a hub and spoke to be able to manipulate and push and pull data from each of the elements and then tie it together at the top. So no, there isn't others that are doing the exact same thing. There are others that do some elements of it, but not to the level that we're more efficient where everybody is tied in and we're interwoven where we're actually able to exploit all the data across the entire community. So no. Thank you. And just one separate question. As far as the vignette you shared, what is the audit process to ensure that what you discovered was accurate after the fact? So a couple of things. So the audit process is really reaching out to other entities, other data scientists within the community, such as folks within the GOINT community. I've got one of the doctors back there. He can tell you and utilize them as the experts in each of their different disciplines to be able to help us make sure that we're doing that correctly. So they help us with that audit process. And there's a community of folks out there that are doing data science. They're just not all collectively together where it's all integrated, which I think about two weeks ago, maybe a month ago, they had a symposium on data science where they brought all these individuals together and they talked about best practices. This Africa Data Science Center was one of those that was briefed at that symposium. I don't know if Gus Wright is here. He may want to elaborate a little bit more. Can we get Gus a mic? Because I do know that Gus was in that symposium also with my data scientists at the same time. Gus, anything? Yeah. I got some, sir. Can you guys hear me okay? To answer your question more, that audit after the fact that here's how machine learning works. A machine learning model is essentially a computer-coded model of the human brain, right? And it has neurons and synapses, but it's written in ones and zeros. So whenever someone makes a discernment and say, hey, here's the answer that this data is telling us to, after the fact you can then go back and correct the model based on the accuracy, then in the future the model becomes a better model, if that makes sense. So that audit and process is circular in so many words. And to answer the colonel's question, tons of collaboration because that deep domain knowledge is such an important asset, you know, it's almost easier to teach when in terms of data literacy, someone who's been doing something like what I do, geospatial sciences or geospatial engineering, it's almost easier to teach me to code than it is to teach me to science. So that's a little bit of correction that we'll have to do in the future to ensure that as we train our soldiers and as we cultivate a paradigm shift and the approach to answering intelligence questions in the future, we'll have to cultivate this skill set as an essential body of knowledge going forward while leaning on experts that have been doing data engineering for the time being, if that makes sense. No, thanks, Gus. That was absolutely spot on. And that's why we need to collaborate with folks like Gus and others within the data science arena. It's a collaborative effort and it's not just one individual. But yes, that's why I say that ours is a silo, but we do collaborate across the IC. JT, you had something? Can you hear me? There we go. Hi, I'm JT Gill, senior civilian at the 207th. So getting to your question, provenance of data, especially when you're talking about unclassified data, like the data that you saw there that's generally drawn from ACLID. And when you use machine learning and natural language, what you're naturally going to get is circular reporting, right? And it may be incorrect right from the start. But the beauty of where it's located within the A's is that we also ever integrate a GWENT detachment, right? And we also have our other intelligence capabilities like our TK on the SIGINT side and then also our HUMAN. And so what we're able to do then is to establish the provenance and the actual progeny of the data to actually get some true ground truth. And ultimately you have to have an analyst to be able to do that. So it's not just putting in data science and generating a computer code because all the computer code's going to do is say this one's like all the rest, right? And if it's like all the rest and it's circular or incorrect from the beginning, you get an incorrect answer. And so really it's about having the analyst there with the data science, scientists to be able to understand the data, be data literate on the analyst side of the house. And then for the data scientists to be able to understand the correct question because ultimately at the end of the day when they write a computer code to go do something, they're asking questions of the data, but if you ask the wrong question, you get the wrong answer. Hey, that's an important point that JT made. Bad data in means bad data out. And so it goes back to what we talked about, what I heard yesterday from Alex Miller, what problem are you trying to solve? And it's not that we've got a data problem and we just have to make sure that we're able to solve it. And this is one of the ways that we get after. But you've got to have the right individuals and the data scientists, data engineers, we absolutely need them. Now earlier I was talking to someone and they asked, well, why don't we just have a bunch of data scientists and start making Intel analysts data scientists? No. I heard you say somebody say yes. No, absolutely not. We need Intel analysts to be experts in the region and understand the region. We don't need more data scientists. You need maybe one or two of those for us. We've got three of them and that suffices. And if we had a hub, we could probably go down to two. But what we need them to do is sit side by side with our analysts and then the analysts who are data literate understand what they're talking about to be able to tell them, this is the outcome that I want to achieve. And then they do the coding and the prototyping and says, OK, this is what I can spit out for you. This is what we can do. Oh, you had 10,000 reports. Well, about 6,000 of those reports were duplicate. We were able to get rid of those. Those are reports that the analyst does not have to read or deal with at that point. Otherwise, if you're doing to, we can't go back to just doing acetates anymore. We can't go back to that and then have the analysts in the back just slopping through reports manually or using their own Excel spreadsheet where they're just manually going through or creating their own pretend code. We've got to have experts who are trained to do that and able to do that in a matter of minutes and sometimes seconds. OK, next question, please. Sir. Hey, sir. This is Chief Chowick. I'll come at you from that side. There are 75th Innovation Command. I used to be supporting 513th MI Brigade. So I feel a lot of this. And I think you guys are on the right track. I have one question for you, though. Who is your data product owners? Is it that you've worn officers in there? No. So the data product owners really are data scientists. They are the ones that curate and they are the ones that are in charge of all the data. Now, it's under the purview of the ace chief who runs the ace. But in terms of, is it the senior worn officer within the ace? No, not at this time. But that's something I'll go back and take a look at. Thanks for bringing that up. Questions, please. Question over here on the right side. Thanks a lot. Sorry, this is a question that's a bit broader than some of the ones that have been asked already. I'm just wondering how you integrate this with other countries such as Australia, especially under the AUKUS agreement. And what sort of challenges you face in doing that? Some of them just bureaucratic. OK, JT, since we talked about that one earlier, I'll let you take that one. OK. That's a fantastic question, right? Because ultimately, when we go to look at a problem, our analysts are asked to look at all the data, right? Not just the data that fits what's releasable to our partners or whatever it may be. That being said, right, depending upon the environment they're in and depending upon our product, we're going to write it to be releasable. So whether or not to be able to put it on StoneGhost or any of the other systems that could be accessible by our partners. One of the ways in which it really gets, especially for our African partners in collaborating with them, is that one of the ways that this translates down is that when our analysts are actually teaching them analysis, probably without even thinking about it, one of the things that they're asking about is, what is the data sources that you have? And then they're already thinking data science type things, like what could you do with this data if I had this data? And because they may not be in our systems and maybe in their systems. And so we're actually teaching our partners to actually be data literate and to think about the possibilities of it. A lot of times, they do not have the advanced tools that our analysts have access to. So for any one given problem of these, there's probably between 20 to 30 different data sets and tools that they're going to use for any specific problem. But those are tools that are already developed. In some of the partner nations where we're actually working, they're not going to have those tools. And so really it's going to be about, OK, how do you think about this data? And then do you have your own integral data scientists to be able to do that? On the larger scale with our foreign partners, that's already happening in terms of our intelligence sharing agreements that already exist. So on the second side of the house, on the geo-inside of the house, some of that is already happening at the 5Is level. At the data science level, that would be more of a mill to mill type arrangement. Does that answer the question? OK. AT, thank you. Hey, I want to thank everybody for your attention. I want to remind you also that what this does, it transforms chaos into clarity. It really is helpful. With a theater as big as we have, a theater as complex as we have, the ability to gain information advantage over our adversaries to think, to have clarity faster than our adversaries can achieve it. This is the secret sauce of the 207th MI Brigade. It's a pilot. And we're looking to learn about it. We're looking to export this around the other MI brigades around the Army. So I just want to thank JT and Mark for sharing this with us. We appreciate your questions and help us learn about it. We're excited about it. We hope you are. Hope you have a great afternoon and enjoy the rest of the AUSA. Thank you.