In modern conflict, the advantage no longer belongs solely to the side with the most sensors or the largest data stores—it belongs to the side that can process, decide, and act faster than its adversary.
General Atomics Integrated Intelligence, Inc. (GA-Intelligence) and General Atomics Aeronautical Systems, Inc. (GA-ASI) successfully completed a milestone demonstration on March 4th, validating the use of Agentic AI to enable autonomous tactical reasoning and decision making beyond line of sight (BLOS) in a sensor Emission Control (EMCON) environment.
In an era defined by information dominance, data interoperability and open architecture are core mission enablers. Across the defense, intelligence, and homeland security communities, the ability to integrate, correlate, and act on data from multiple domains is central to achieving decision superiority.
At GA Intelligence, we're constantly exploring the boundaries of what modern AI and large language models (LLMs) can achieve. In this blog post, we'll explore the general mechanisms used by AI and LLMs and how they relate to solving games and puzzles.
In light of some recent developments and announcements regarding secure systems needed to support our warfighting information technology requirements (or the lack thereof), we at GA Intelligence have some things we’d like to offer up for consideration and discussion.
General Atomics Aeronautical Systems, Inc. (GA-ASI) and General Atomics Integrated Intelligence, Inc. (GA-Intelligence) successfully completed a groundbreaking demonstration on July 8, integrating technologies across multiple affiliates to showcase long-range kill chain effects, including an autonomous air-to-air engagement.
The Panama Canal is a critical pathway for global trade which facilitates billions of U.S dollars’ worth of trade annually. Despite incurring greater transit fees than alternatives, such as the Suez Canal, the Panama Canal remains the preferred shipping route for Southeast Asian goods destined for the U.S. East Coast because it reduces cargo time in transit by an average of six days.
Artificial Neural Networks (ANN) are universal function approximators with layers of feedforward computational nodes. ANNs are used in many Data Science applications involving classification and regression.
GA-Intelligence Optix platform provides easy access to low latency, high volume geospatial data and analytics at scale.
At GA-Intelligence our primary goal is to extract useful information from the data we process. We’re constantly striving to improve our data processing tools to enable powerful workflows and easier development.
Change detection from aerial or satellite images is in great demand for monitoring various activities. The detection of wide area change over the past year has garnered a lot of interest, with a wave of publications using different methodologies and different data sources to solve the task [1,2].
The increase in Radio Frequency (RF) devices driven by the Internet of Things (IoT) has led to a need to efficiently and reliably authenticate these identifiers, especially since traditional authentication methods are easily compromised through identity spoofing.
Image classification is used on overhead imagery to detect specific objects such as planes, cars and buildings.
Algorithms for effective automated target recognition (ATR) at scale must be able to process large imagery datasets in a timely manner, while minimizing human effort in terms of annotating images and reviewing model predictions. Convolutional neural networks (CNNs) excel at many computer vision tasks, but can have a high rate of false positives under certain conditions.
a machine learning library comparable to scikit-learn. This library uses enumerated encoding to store categorical variables. With enumerated encoding, each variable is mapped to an integer in one singular new feature.
With the global pandemic disrupting supply chains everywhere, we at GA-Intelligence were interested to investigate the reliability of shipping schedules. To do this, we partnered with global carrier aggregator Linescape.
Every day, hundreds of thousands of vessels cross the oceans, traveling from country to country and port to port. Most of these trips are mundane, routine, and fairly uninteresting. These vessels are carrying cargo and have declared their port calls beforehand.
Modern computer vision models often do a fantastic job of identifying the contents of an image, but they can also make mistakes, confusing one type of object with another. These mistakes sometimes make human-interpretable sense, but other times the reason for the mistake is less than obvious.
Overhead imagery is widely used in a variety of applications, from monitoring natural disasters to visualizing directions on a map. For example, Google Earth uses overhead satellite imagery covering nearly the entire Earth’s surface.
A recent article on the U.S. Army Corps of Engineers ERDC (Engineer Research and Development Center) website described how a patent was granted in September 2020 to a system invented by GA-Intelligence Michelle Hamilton for Multi-Criteria Decision Analysis.
There are many reasons that researchers analyze maritime traffic. For example, commodity traders analyze the flow of goods, port operators coordinate ground transportation and minimize vessel time in port, and government agencies track economic statistics.
The GeoMesa team has released LocationTech GeoMesa 3.1.0. This minor release adds new exciting features and upgrading from GeoMesa 3.0.0 should be easy.
According to GA-IntelligenceEthical Principles, individual privacy is a right. When statistical microdata (data about individuals) is released to the public or other non-trusted parties, it must go through a process of anonymization to protect the privacy of the individuals represented.
Technology follows a common story arc. First, technology is created that makes it possible to do something that was previously impossible. Next, incremental improvements to that technology improve the performance of that task in a number of ways.
Part of being a successful data scientist is the ability to to clearly visualize your data and results in an easily interpretable way. This is not always a simple task.
GA-Intelligence helps our customers make big decisions, often within a large and complex ethical context. We have recently published our Statement of Ethical Principles, which provides a framework for thinking through ethical problems as part of our design and development.
At first glance, this problem seems trivial. Neural nets are sophisticated technical constructs capable of advanced feats of machine learning, and you learned the quadratic formula in middle school. But an interesting property of classifiers was revealed trying to solve this issue.
We recently celebrated the 30th anniversary of the founding of GA-Intelligence, and to celebrate that occasion and give back to our local community, we sponsored an internal hackathon for our staff one recent afternoon.
Self-driving vehicles can read road signs and identify pedestrians. State-of-the-art diagnostic tools can evaluate MRI and CT scans to detect illnesses. Your smartphone can search through your photos and pick out all of the ones with cats.
The winter 2019-2020 issue of c-biz magazine features a cover story titled “Workplace Wellness” that described employee wellness options at several forward-thinking Charlottesville area employers.
We are always excited to see customers using our technology to advance the common good. The Big Data UN Global Working Group is leveraging CCRi’s Optix platform to deploy a multidomain analytics platform with the goal of generating national statistics.
Rather than simply flying straight from point A to point B, commercial airline flights fly along preset routes. Pilots often liken these routes to “highways in the sky.” However, unlike highways on the ground, exact maps of these routes are hard to come by.
One benefit of working at GA-Intelligence is the opportunity to stretch outside of our comfort zone. We consider developers to be innovators and solution engineers who can tackle any problem presented with the skillset we are continually building.
With the increasing availability of network connected GPS devices, aircraft trajectory data is now quite common. Mining trajectory data for patterns presents a number of unique challenges:
It’s been a busy winter here at GA-Intelligence with expansion of all types going on. We are growing with new and expanded projects, our operations practices are getting fine-tuned, and we continue to grow physically here at our Sachem Village headquarters, constructing additional space in our newest building.
Representing entities by their attributes, like just about any database does, makes it easy to manipulate information about these entities using their attribute values—for example, to list all the employees who have “Shipping” as their “Department” value.
Summer finally wound down and crept very slowly into fall. And, boy, so far it’s been a wet one! The majority of days in September experienced everything from mild mist to heavy downfalls as one storm system after another moved through our area.
The image below loops an animation of Optix.Earth’s rendition of ship traffic around the southeastern United States from Tuesday to Thursday of last week. It shows that all the ships are avoiding an area about 800 miles wide that was moving towards the North and South Carolina coast: Hurricane Florence.
This past June GA-Intelligence celebrated a big birthday of turning 29 years old! We’ve done a lot of growing up since the company started out as a little three person start-up out of UVA, and arriving at our current location in Sachem Village by the year 2000.
The International Society for Optical Engineering, or SPIE (formerly known as the Society of Photographic Instrumentation Engineers), has published a range of peer-reviewed scientific journals for over 50 years.
TransE, or Translating Embeddings for Modeling Multi-relational Data, lets us embed the contents of a knowledge graph by assigning vectors to nodes and edge types (a.k.a. predicates) and, for each subject-predicate-object triple, minimizing the distance between the object vector and the translation of the subject vector along the predicate vector.
What’s the first thing that comes to your mind when you think of the month of May? Flowers? Parades? Well, if you’re an avid cyclist (and have Frank D in your employ) then the correct answer is The National Bike Challenge! CCRi cyclists participate in this event each year with great enthusiasm and good-natured competitiveness.
GeoMesa, like many other data stores that index geographic data, uses space-filling curves to impose an order on two-dimensional geometries. It’s easy to know that “Virginia” as a string data type follows “Illinois” when sorted alphabetically, but it’s much less obvious whether the multi-polygon that is Virginia’s border should come before or after Illinois’ state boundary in an index.
The GeoMesa project has just published our official 2.0.0 release. This is a quick highlight of some of the new features. For a full list of the improvements and fixes, take a look at the release notes.
Ah, late winter and love is in the air! Or maybe it’s just the early start to pollen season…? Valentine’s Day reared its ugly/awkward head so naturally a special symposium was in order—Valsposium! The usual snacks were in play, along with other appropriately theme treats.
A number of healthcare metrics, including readmission rates and chronic disease burden, have been linked to socioeconomic indexes in various studies. In the United States, many of these indexes can be calculated from freely available US Census Bureau data. Generally, the index is a simple function of one or more census variables.
In the current Big Data environment, it’s crucial to have a quick iteration cycle for exploring data, performing analytics, and visualizing the result. For example, the impact of Hurricane Harvey on oil and gas shipping in the Gulf of Mexico last fall has many dimensions to explore...
Ontologies are widely used for representing and reasoning about semantic content in a structured way. However, manual ontology construction is a subtle and time-consuming process that often yields mixed results. Hand-crafted ontologies tend to be inflexible and inordinately complex, which limits their usefulness and makes cross-domain alignment painfully difficult.
Active imaginations and a love of fun are always in play at GA-Intelligence . Leave it to Andrew H. to come up with an idea so brilliant it left us shaking our heads thinking, “Why didn’t we think of this before?”
One of the most interesting areas where we’ve seen usage of GA-Intelligence open source spatio-temporal database GeoMesa has been in precision agriculture.
Oh, what an activity-filled summer and fall we’ve had here at GA-Intelligence! Count on us to have fun in a variety of different ways. We’ve had dinners and drinks downtown, we’ve sampled a new brewery, and seen a total eclipse of the sun.
or several years now, various GA-Intelligence employees have attended Charlottesville’s pop-up tech conference becamp, and this year we took the additional step of becoming a sponsor.
According to the Wikipedia page for the game Battleship, the Milton Bradley board game has been around since 1967, but it has roots in games dating back to the early 20th century.
GA-Intelligence annual spring brewfest once again put beer tasters in the difficult quandary of trying to rank favorites from the plethora of home-brewed goodness. While the kids played in large, muddy ponds from recent rains, the brew tasters thoughtfully sampled, and then sampled again.
A great deal of machine learning progress in recent years has been based on the training of neural networks using supervised learning.
Last summer, our blog post “GeoMesa analytics in a Jupyter notebook“ described how Jupyter Notebook allows interactive exploration of data using programming languages that are rarely used interactively. It also showed how a series of steps can be saved in a reusable notebook for others to learn from.
Sometimes you just need to take a break and enjoy the little things in life. Good coffee. Good beer. Good food with friendly co-workers. Better yet, Good food, good beer, and some tough trivia.
Here at GA-Intelligence, we’re working on such an interesting set of technologies that we could probably find a conference to speak at every week if we wanted to. Two especially nice conferences that recently included CCRi speakers were the Google Cloud Next ‘17 conference and Charlottesville’s Tom Tom Festival Machine Learning conference.
Release 1.3 of GeoMesa has taken some great steps in making GeoMesa an even better analytics platform for Big Spatial Data. Many improvements fall into two categories: support for a wider range of Spark features and improved support for interactive notebooks such as Jupyter and Zeppelin.
One of the more interesting research projects underway at GA-Intelligence uses machine learning to automatically generate text descriptions of action happening in videos, complete with a color-coded playback slider showing where in the video the action takes place.
As described on their About page, “GeoHipster is a collaborative independent online publication on geotechnology and culture, with special focus on open source and open data.” It’s a great way to stay in touch with interesting new things in the geospatial world.
It’s a cliché that a picture is worth a thousand words, but what if you want to create a picture of two thousand nodes connected to each other with several fields of information attached to each node?
For an ongoing project at GA-Intelligence, we wanted to determine whether remaining with Torch (used for Phase I of a project currently underway at GA-I3 running on GPUs) or switching to TensorFlow or Theano made the most sense for Phase II of the project.
An article in last month’s Wired magazine titled “The A.I. Enigma: Let’s Shine a Light into the Black Box” (not available online) described how the inscrutable nature of many artificial intelligence algorithms has frustrated people who want to know why a system made a particular recommendation.
As the final muggy days of summer rolled to a close, a kind of frenzy broke out among some denizens of GA-Intelligence. A frenzy that could only be quenched by crispy green vegetables or exotic local fruits.
When starting a new data science project, one should typically first understand the problem domain, state a hypothesis, and then gather data to prove or disprove said hypothesis. But, if I’m being honest, I usually start with a new data set and ask “What kinds of problems can I solve with all this new data?!” I can’t help myself—a new data set is too intriguing to resist diving into.
It’s been another busy summer for GA-Intelligence. We expanded our teams into two new buildings: one small office building and into a larger office building, all within easy walking distance, within the same office park, for a total of four buildings.
Earlier this summer, the GA-Intelligence blog entry GeoMesa analytics in a Jupyter notebook introduced the use of Jupyter notebooks with Scala and GeoMesa to do Apache Spark analytics and geospatial visualization. As the blog entry said, it only scratched the surface of the kinds of geospatial analytics that you can do with GeoMesa and Spark.
We asked each of our interns to describe what they’ve been working on and how they were enjoying their life in the working world. This year, all the interns came from the University of Virginia School of Engineering and Applied Science, where all are Computer Science majors except for Ryan, who is majoring in Biomedical Engineering and Physics.
As described on its home page, “The Jupyter Notebook is a web application that allows you to create and share documents that contain live code, equations, visualizations and explanatory text. Uses include: data cleaning and transformation, numerical simulation, statistical modeling, machine learning and much more.”
Over the past couple of years GA-Intelligence has experienced incredible growth. So much so that our employee population has more than doubled. Add family and SO’s, and you have the makings for quite a big turn out under our trusty Shelter #2 in McIntire Park. And after weeks of daily rain fall, Mother Nature cut us some slack and came through with sunny skies and gloriously perfect temperatures.
One great new feature of GeoMesa 1.2 is its prototype support for Apache Cassandra. This open-source key-value NoSQL database was originally developed at Facebook to let users search their inboxes more efficiently than the MySQL database that Facebook had been using.
When not at work, GA-Intelligence employees are actively at play. Many engage in competitive or recreational athletic pursuits like tennis, cycling, soccer or Ultimate Frisbee. Those with children are often on the sidelines, encouraging their kids at the soccer field or swim meet.
Well-known GeoMesa use cases often center around its ability to work with large, scalable column-oriented database managers such as Accumulo, HBase, and Google Cloud Bigtable.
Release 1.2 of GeoMesa, our open source suite of geospatial analytics tools designed to run with Hadoop-scale volumes of data, is now out.
Albemarle County’s GA-Intelligence named a Bronze Level Bicycle Friendly BusinessSM by the League of American Bicyclists
Sunny skies and perfect fall weather were the backdrop for GA-Intelligence Oktoberfest in McIntire Park. And while there was the usual grilling, great potluck, games and camaraderie, there was also a palpable feel of competition in the air.
Sometime at the beginning of this summer, Frank Deviney told me about the National Bike Challenge, but his words fell upon deaf ears. Frank had been logging his miles for many weeks by mid August with his e-bike (which is allowed by the challenge) when I decided I could use the motivation to cycle and decided to register.
Sometimes Data Scientists find themselves with a map-reduce cloud architecture and computation that needs to be done on a large scale, but the data isn’t actually cloud scale.
In my previous post, I laid out a case for the significance of Big Spatial Data and my surprise that so little is published about it. Now I’ll formally introduce the GeoMesa Project, which seeks to address the challenges faced by developers buried in data with no way to use it.
Despite all the buzz that surrounds the term “big data,” we hear surprisingly little about “big spatial data.” Of the top ten results from an Incognito Google search for “Big Spatial Data,” only two results are from 2015 and only one – “GIS Tools for Hadoop by ESRI” – attempts to present a solution.
Modeling language with neural networks is a highly active and rapidly maturing field. Language agnostic models can now be efficiently trained on raw text containing billions of words, and they achieve state of the art performance on many natural language processing tasks such as tagging parts of speech, semantic role labeling, and sentiment analysis.
GA-Intelligence in collaboration with Google, Inc., has announced the initial release of GeoMesa for Google Cloud Bigtable. GeoMesa is an open-source system that quickly stores, indexes, and queries hundreds of billions of geospatial features in a distributed (i.e. cloud) database.
Spring is in the air, and once again it’s time for our semi-annual Brew Fest and Picnic in McIntire Park. But wait! The clouds are darkening, and the threat of rain looms large. No matter. GA-Intelligence employees are quick to adapt to change when needed.
Partial function application is the process of fixing a number of arguments to a function. This is an interesting and useful feature in functional programming languages such as Scala, R, and Javascript.
Recently at GA-Intelligence, we have been doing a lot of research in the area of reduced dimensional semantic embedding models: models where semantically similar objects possess similar representations.
I had the privilege of presenting my work on “Calculating Feature Importance in Data Streams with Concept Drift using Online Random Forest” at IEEE Big 2014 in Washington, DC this last week.
Grey skies and persistent drizzle did nothing to deter the GA-Intelligence faithful from participating in their traditional Ocktoberfest in McIntire Park.
At GA-Intelligence we are interested in all kinds of learning algorithms, from generalized linear models to random forests. One of the models that we have recently been working on is called a rule ensemble and aims to provide deeper insight into how a model produces a prediction while still maintaining predictive accuracy.
This blog was written by each intern, sharing what they did this summer as well as interesting parts of their GA-Intelligence internship experience. As you can see, our software engineering interns really dove into the code and had a productive summer!
Learning from relational data, in which entities can be connected to each other with varying types of relations (friendOf, worksFor, …), frustrates many machine learning algorithms which expect tabular data. How do you represent all the friends of an individual in a table?
As a longtime bicyclist, Alec Gosse is concerned with bike safety and the desire to make bicycle travel practical in a society centered on automobile travel.
Learning from relational data, in which entities can be connected to each other with varying types of relations (friendOf, worksFor, …), frustrates many machine learning algorithms which expect tabular data. How do you represent all the friends of an individual in a table?
GA-Intelligence is not only a fun place to work, it’s also a great place to have fun. One of the perks is enjoying the fermented treats of our creative brewers. Hunter Provyn brought in a keg of his own “Childe of Stingo,” the offspring of the classic English strong ale, brewed with oak chips and partially soured.
GeoMesa is an open-source, LocationTech project that manages big geo-time data within the Accumulo key-value data store so that those data can be indexed and queried at scale effectively.
In this and my next post, I’ll be showing a a few quick analyses we performed using a new tool we developed, called Elias. In today’s post, we’ll see how topic modeling can be used to characterize how entities are co-mentioned, not just how often.
GA-Intelligence was delighted to host the second meeting of the Cville Data Science group earlier this month. A full house packed our conference room, and a good time was had by all.
Here at GA-Intelligence, we do a lot of machine learning; going from data to knowledge is kind of our thing. We’ve got a library of machine learning tools in-house, and we even use other algorithms from third-party developers sometimes.
When using actors, sometimes a parent actor needs to aggregate a result from a list of child actors. And if the parent doesn’t want to wait around forever for the children to get their responses together, it might hand off the responsibility to a temporary actor.
When building web based data visualization applications, you usually want to defer rendering and styling to as late as possible in the request lifecycle. This gives your users the most flexibility when composing and customizing intelligent visualizations of data.
Alexandria, VA October 27, 2013 — Boundless and GA-Intelligence announce the launch of a new platform for building geospatial capabilities on the highly distributed Apache Accumulo database. OpenGeo Suite with GeoMesa combines data management and publishing with big data analytics via GeoMesa, a high performance spatio-temporal indexing and querying capability for Accumulo.
Scala lets us write short and expressive code, and today we’re going to look at one of the ways we’ve leveraged that ability in the GeoMesa codebase.
Big Data has driven the need for datastores that can scale horizontally leading to the development of many different NoSQL database implementations, each with different persistence and query philosophies.
