The decision on the table
Fraud rings act together. Most transaction-risk models score accounts and pairs of accounts, which adds up individual signals and can miss behaviour that only appears when a group moves at once.
Finance · financial crime
Screen for the coordinated fraud rings pairwise models miss
A method for reading collective behaviour in payment networks. We measured where pairwise models go wrong; no detection gain is claimed.
Headline result
72% pairwise error found
The pairwise prediction was wrong by 72% on the collective response we measured
Measured on quantum hardwareClassical computation
What we did, with proprietary methods
We treat the payment network as a coupled system and read its collective response, on quantum hardware and in simulation, and compare that with what a pairwise-additive model predicts for the same network.
What it was worth
The pairwise prediction was wrong by 72% on the collective response we measured
See where your pairwise model is blind
- Baseline
- A pairwise-additive prediction of the same network
- Condition
- The pairwise prediction was wrong by 72% on the collective response we measured
What happened next
The 72% blind spot in the pairwise model is a method result. The stated next step is a detection pilot on the client's own data; no detection gain is claimed before it.
Decision it enabled
See where your pairwise model is blind
Instead of: A pairwise-additive prediction of the same network
Verified by · Device result + simulation, checked against exact references.
How we know
- 01Pairwise-additive prediction error of 72% against the measured collective response, validated against exact references.
- 02Not claimed: a fraud-detection gain. This is a method result; a detection pilot on your data is the next step.
StatusMeasured on quantum hardwareClassical computationMethod measured on quantum hardware and in simulation. No detection gain claimed.
Next story · Mobility · aerospace
A dated roadmap for aerodynamic flow simulation
Semiconductors · materials
Pick the winning EUV photoresist before the fab
Resist candidates ranked before any were made, so only the lead goes to synthesis and wafer testing.
Measured on quantum hardwareClassical computationRead the story
Within 1.4% of exactPharma · life sciences
Find where a protein can be drugged before the assay
Candidate allosteric sites read from a structure file alone, with the site model validated against exact references. A method result: no drug-ranking gain is claimed.
Measured on quantum hardwareClassical computationRead the story
Mobility · autonomy
Sign off an autonomy controller as safe against real disturbances
A disturbance model measured on a 128-qubit processor was the only one that certified a controller safe at its operating point.
Measured on quantum hardwareClassical computationRead the storyDecide before you build.
Bring us the decision you have to make next.
We tell you which option wins, and whether quantum is the reason.
No keyNo cardAn answer in about a second