Sharing financial crime knowledge

Patent metadata

Field	Value
Publication number	US-11954174-B2
Application number	US-202017063731-A
Country	US
Kind code	B2
Filing date	Oct 6, 2020
Priority date	Oct 6, 2020
Publication date	Apr 9, 2024
Grant date	Apr 9, 2024

Abstract

A computerized-method for scaling automatic deployment of a machine-learning detection model in a cloud-based managed analytics service by knowledge sharing to overcome an imbalanced dataset learning problem. The computerized-method includes: sending the received data to machine-learning models to synthesize patterns of the received data to yield a differential privacy data; maintaining in the database the differential privacy data of one or more on-prem cloud-based managed analytics services to generate a consortium shared synthetic data lake; operating phases of machine-learning detection model based on the received data and data in the database to create a packaged model. The data in the database is aggregated and used during the operating phases of the machine-learning detection model to create a packaged model for other on-prem cloud-based managed analytics services, thus overcoming imbalanced dataset learning thereof, and after the packaged model is created it is automatically deployed on-prem.

First claim

What is claimed:

1. A computerized-method for scaling automatic deployment of a machine learning detection model in a cloud-based managed analytics service by knowledge sharing, to overcome an imbalanced dataset learning problem, said computerized-method comprising:
a computerized system comprising a processor, a database, and a memory;
a plurality of client computers each comprising an on-prem cloud-based managed analytics service;
wherein the on-prem cloud-based managed analytics services continuously monitor criminal typologies and operating fraud and Anti-Money Laundering (AML) model performance;

a storage comprising a consortium shared synthetic data lake,
wherein the storage is remote from the computerized system and the plurality of client computers,
wherein the consortium shared synthetic data lake comprises differential privacy data received from the computerized system,
wherein the differential privacy data excludes any individual personal details,

the computerized system:
receiving financial transaction data from an associated on-prem cloud-based managed analytics service,
determining fraudulent financial transaction data in the received financial transaction data;
generating, by machine learning models at the computerized system, new differential privacy data by synthesizing patterns of the fraudulent financial transaction data;
wherein the new differential privacy data excludes any individual personal details,
wherein the synthesized patterns are at least one of: fraud patterns or AML patterns;

storing the new differential privacy data in the database of the computerized system;
updating the consortium shared synthetic data lake by adding the new differential privacy data to the consortium shared synthetic data lake;
creating a packaged machine learning detection model, by the processor of the computerized system, based on (i) the received financial transaction data and (ii) data obtained from the updated consortium shared synthetic data lake,
wherein the data obtained from the updated consortium shared synthetic data lake comprises differential privacy data associated with financial transaction data from at least one different client;

deploying the packaged machine learning detection model on-prem of the client of the associated on-prem cloud-based managed analytics service.

2. The computerized-method of claim 1 , wherein the synthesized patterns are related to financial crime.

3. The computerized method of claim 1 , wherein the machine learning models to synthesize the patterns are Conditional Table Generative Adversarial Networks (CTGAN).

4. The computerized method of claim 1 , wherein creating the packaged the machine learning detection model includes the steps of: (i) data validation; (ii) data preprocessing: (iii) data enrichment; (iv) data sampling; (v) feature engineering; (vi) model tuning and training; (vii) model testing; and (viii) model packaging.

5. The computerized-method of claim 4 , wherein the data obtained from the updated consortium shared synthetic data lake is used in the data enrichment phase.

Assignees

• Actimize Ltd

Inventors

• Pati Debabrata
• Dahiphale Pravin
• Butvinik Danny

Classifications

• G06N3/0475

  Generative networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/094

  Adversarial learning · CPC title

• G06F18/2148 Primary

  characterised by the process organisation or structure, e.g. boosting cascade · CPC title

• G06F16/254

  Extract, transform and load [ETL] procedures, e.g. ETL data flows in data warehouses · CPC title