Flashback has developed a state-of-the-art high speed patent-pending software technology called CipherSensor™, which enables active, long term learning from extremely large (terabytes), continually changing data-sets enabling more accurate estimation, prediction and control of desired results.  Originally developed by one of its founders, Greg Grudic, PhD, as part of an autonomous, image-based robot navigation system (under the DARPA-sponsored Learning Applied to Ground Robots (LAGR) program), CipherSensor's competitively novel, interpretive and predictive analytical approach yields opportunities to produce groundbreaking data analysis solutions across a variety of industries. CipherSensor attributes include:

  • Machine learning algorithms based on computer vision, statistics and robotics research
  • Automated, real time feature extraction
  • Capable of analyzing extremely large volumes of continuous data generated from multiple sources
  • Leads to identification of application critical information and improved estimation/prediction capabilities
  • Adapts to new information ensuring the model utilized is accurate for the information it is analyzing
  • Identifies when models are not applicable, preventing false classifications
  • Enables early and timely decision making
  • Supports development of fully autonomous systems
  • Increases potential for improved outcomes at a lower cost

Most current data analytic models and algorithms, including those with statistical and machine learning origins, rely on a priori knowledge about the data streams of interest: their current uses as metrics for measuring certain conditions, as well as their relationships to one another. These requirements often bias subsequent analyses and make it difficult to provide accurate predictions and could result in ignoring relationships that were not considered consequential at the outset. In contrast, the CipherSensor Technology analyzes these large data streams without any knowledge of their origin or potential role, but rather in the context of the desired outcomes or criteria of success in which the data streams exist. The CipherSensor computational engine consumes massive amounts of information, determines the relevant information in the dataset based on its relationship to the outcome of interest, and utilizes that information to create predictions of how likely the outcome of interest is at any given time.