Latest News
EU AI Act Mandates Understandable AI for all High-Risk AI SystemsLeading Corporations Adopt New M&A Due Diligence Criteria to Assess Value, Risks of Data-, AI-Centric BusinessesDiveplane and Cantellus Group Announce Partnership to Promote Adoption of Understandable AI®
Skip to main content

Find, understand and resolve the anomalies in your data and models.

Our patented technology finds outliers where humans can’t. SONAR™ performs forensic analysis of data and models to ensure anomalies are tracked, identified and their source understood so action can be taken quickly.

Model drift management is now essential functional requirement for any organization deploying machine learning capabilities. Model drift may occur quickly or over time but the results can be devastating and impact the confidence a business can have in their business intelligence strategies. SONAR™ delivers continued model monitoring to detect drift and deviations allowing the user to understand exactly what data is impacting the model and that take the necessary to optimize the model performance

SONAR™ is our anomaly detection tool which uses our patented technology to identify outliers and the more insidious “inliers” (outliers that exist within close proximity to real data points) in a dataset and helps to identify data anomalies and model drift before they become a real problem.
Diveplane SONAR
Diveplane SONAR

The Deterrent Dividend

SONAR™ is unique. While other AI engines are black boxes that don’t show the work behind their answers, SONAR™ answers are understandable and auditable, critical to building trust and maintaining proper accountability in regulated industries.

Fraud, Waste, and Abuse (FWA) is a scourge that negatively impacts corporate earnings and market efficiency:

The National Heath Care Anti-Fraud Association estimates that US health insurers lose $68 billion annually to fraud, equal to 3 percent of the nation’s $2.26 trillion in health care spending.

As little as 5% of what is lost to fraud is ever recovered.  By 2026, when healthcare costs are projected to be in excess of $5.5 trillion, this may mean losing $165 billion to fraud, waste, and abuse (FWA) every year.

According to a 2018 US Department of the Treasury estimate, domestic financial crime (excluding tax evasion) generates approximately $300 billion of illegal proceeds annually.

Furthermore, the “many varieties of fraud, including bank fraud, consumer fraud, healthcare fraud, securities fraud, and tax refund fraud are believed to generate the largest share of illicit proceeds.”

In addition to the normal cost of fraud, banks and other financial institutions must also incur the cost of the compliance efforts associated with detecting and reporting suspicious account activity.

These can be indicative of money laundering operations that “wash” illegal revenues generated by fraud, drug trafficking, human smuggling, human trafficking, organized crime, and corruption.

By deploying Diveplane SONAR™, Companies and Organizations have the potential to gain the Deterrent Dividend, the anticipated downturn in FWA by declaring the use of SONAR™ as part of its FWA reduction strategy.

Benefits of SONAR™ include:

Finds anomalies in large, complex datasets and ML models.

Flags suspicious cases objectively

Decreases level of manual intervention

Complements the work of data scientists, auditors and investigators

Learns from expert feedback

Business Benefits of SONAR™

SONAR™ can be implemented for online learning to monitor and detect data anomalies and model drift in real time, giving the user immediate feedback to identify underlying shifts in data distribution and situations where the model may no longer have predictive power. From the SONAR™ feedback the user can then further do a root cause analysis to identify:

  • changes in data (e.g. anomalous patient population)
  • model no longer accurate (e.g. change in diagnosis method)
  • changes in feature importance (if accuracy remains steady, the original model may be making the correct predictions but for the wrong reason)

Special Investigators and auditors are tasked with the unenviable job of identifying FWA or money laundering cases by hand. This involves painstaking interrogation of data to identify fraudulent transactions. Given the arduous nature of the process, it’s not surprising that the rate of recovery is such a small proportion of the estimated value of annual FWA in the United States.

Whether the result of genuine error, opportunism, or professional conspiracy, SONAR™ significantly increases the amount of data that investigators can analyze. SONAR™ detects the anomalies that set suspicious financial transactions apart from the rest and flags them for further examination.

SONAR™ Deployment

Diveplane has a number of flexible deployment methods to suit every customer requirement, from on-premise to public or private cloud.

Our Professional Services Team will provide training and ongoing support to all our customers to ensure they achieve the maximum benefit from the software.