What We Do
Artificial Intelligence and Synthetic Data
Synthetic Data can train and test artificial intelligence applications at a much quicker pace compared to its non-artificial data counterpart. Threat detection, machine learning algorithms, and behavioral pattern identification are among some of the many ways synthetic data can help progress the advancements of artificial intelligence; because testing and training models using synthetic data is also more flexible than when using authentic data, the technology we develop can go through a much more extensive amount of troubleshooting and still be developed at a faster rate.
Census & Statistics
Simulated Census Data Environment
The Census Bureau’s major challenge for Decennial Census 2020 is to successfully develop and deliver a new highly integrated complex system that captures data from multiple modes including web, mobile, handled devices of enumerators, telephony and paper, on time and within budget. Smart Data will enable the Bureau of Census (BOC) to have multiple opportunities to test and get this right before Census Day by speeding up development and reducing risk. The BOC will be able to run a simulated Census before the actual 2020 Census by using high fidelity (data that mirrors real data in complexity) production scale test. Smart Data will simulate components of the entire census operation, including associated metadata and paradata, with the attributes of realism, scale, scenario ground truth (data that is generated for each dataset and is precisely known by field) and complex interconnectivity. High precision measurements of the entire integrated census system environment’s’ accuracy and performance, to enable efficient and effective root cause analysis, will now be possible.
ExactData provides US Federal Agencies with Smart Data for tax return and healthcare data. We manufacture Smart Data spanning Census, Tax, Healthcare, and others, maintaining internal and longitudinal consistency across the comprehensive sets of data.
Public Accessing Website
Smart Data can act as a script for automated entry through a web application with correct request/response pairing tags for integration with Virtualized Services Tools.
Banking applications and credit card systems can be tested profusely with synthetic data before software is rolled out to live servers, and with our smart data we can simulate live consumer interactions with minimal risk for all parties involved.
Today's cyber security solutions are complex, interdependent architectures that look at a wide variety of signals from a huge number of heterogeneous components. While it is possible to independently test each component of these cyber architectures, it is increasingly challenging to test the interactions between these components. Attackers attempt to exploit the seams in these defenses, knowing that complex systems interact in complex and often unknown ways.
Attempts to test these complex systems usually consist of using network traffic generators to send a variety of protocols and files into the network or using tools to emulate various user and system behaviors. The challenge is that these systems bring very little sophistication to their work. The "background" traffic is often repeated characters, simple packets, or repeated, simplistic behaviors. Many of the systems to be tested can very easily discard most of the test traffic, leaving effectively full compute resources to search for the simulated attack traffic or malicious user behaviors. This can overstate system effectiveness in the real world. Another approach is to collect production network traffic and user actions and then attempt to sanitize those for test purposes. However, there is a huge risk of leaving personally identifying information, otherwise known as PII, or company intellectual property in these sets. Further, the types of protocols and behaviors that can be captured depend mainly on chance and timing.
ExactData approaches this problem in a very different way. Working closely with our customers, we develop a data model that captures the complexity of their desired environment. This allows us to generate a large amount of internally consistent test data that covers every desired behavior. For instance, to generate user activity inside an Enterprise, the data model can be populated with thousands of test employees, new hires, terminating employees, contractors and other third-party users. The data model can include a variety of endpoints, servers, and network devices. When the data model is populated, it can then be run over any desired time frame and is capable of many kinds of different output such as log-on and logoff activity, server or network resource access, emails to and from colleagues and business partners, and inbound and outbound HTTP sessions with well-formed artifacts.
The ExactData approach means that the generated cyber testing data set:
All while having absolutely no risk of containing PII or intellectual property!
Relevant development databases including future state on demand will decrease your development cycles and time while increasing the quality of your deployed systems with fewer errors to correct downstream.
Fraud use cases are designed into the data models along with expected systems response files. Smart Data will allow you to measure your fraud system for Accuracy, Precision, Capture Rate, Escape Rate, False Negatives and False Positives.
Collaborative Industry Sandbox
Smart Data to run non-confidential challenges to evaluate best of breed technologies to address confidential needs
Threat Detection use cases are designed into the data models along with expected systems response files. Smart Data will allow you to measure your fraud system for Accuracy, Precision, Capture Rate, Escape Rate, False Negatives and False Positives. Ask about our project with DARPA to develop insider threat detection algorithms using billions of records simulating a corporation’s digital signature over a 10-year period interwoven with sophisticated threat patterns
Simulation & Modeling
Run your simulation and modeling scenarios against large Smart Databases that will accurately measure the performance of both the human and information technology components of your system.
Software & Interface Development
ExactData has made practical the ability to generate relevant test data, on demand, in virtually unlimited quantities. This manufactured correlated Smart Data is automatically ingestible into legacy systems or formatted for automatic ingest into systems including the use of HL7 messaging formats. Interfaces can now be development and tested in this simulated correlated data environment, which includes not only the legacy systems, but also the future state modernize system.
Relevant data sources are needed to test and evaluate tools that can be hosted in a cloud environment and accessed by personnel with no privacy risks. ExactData can deliver these relevant databases and automatically post to the cloud environment. The solution development and deployment process would be accelerated by having development work done against the simulated data without the need to access production facilities, submit for clearances nor provide access to confidential production databases. Smart Data would enable you to “try before you buy” enabling better acquisition decisions and faster technology deployments across the Enterprise.
Technology Partner Integration and Certification
ExactData can provide relevant bases for development of interface software with your technology partners and certify compliance. This will increase your partner network, with a direct positive impact on revenue.