# US 2014/0156568 A1 · Self learning adaptive modeling system

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2014/0156568 A1
- Document ID: US20140156568A1
- Status: Published application
- Kind: A1
- Application number: 13/706,318
- Filed: 2012-12-05
- Published: 2014-06-05
- Earliest claimed: 2012-12-05
- Assignees: Microsoft Corporation
- Applicants: Microsoft Corporation
- Inventors: Sandipan Ganguly; Lu Xia; Weiwei Wu; Shoou-Jiun Wang; Justin Hobart
- Web page: https://patentgrep.com/patent/US20140156568A1

## CPC

- G06Q 10/0637: Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06Q 10/06375: {Prediction of business process outcome or impact based on a proposed change}
- G06Q 10/067: Enterprise or organisation modelling
- G06F 18/214: Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F 18/2411: based on the proximity to a decision surface, e.g. support vector machines
- G06N 5/025: Extracting rules from data
- G06N 5/046: Forward inferencing; Production systems
- G06N 5/048: Fuzzy inferencing
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/00: Machine learning
- G06N 20/10: using kernel methods, e.g. support vector machines [SVM]
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Self-learning and adaptive modeling is employed with respect to predictive analytics. A hierarchical model structure can be employed comprising a set of predictive models automatically built from accumulated data and distributed across multiple levels. For a given input type, a set of candidate models can be identified across varying levels of granularity, and a best model selected based on a comparison of performance metrics of the models. The best model can then be activated for use in making predictions. Of course, the best model can change based on most recent training performance results, since as more data becomes available more specific models can be developed.

## Claims

### Claim 1

A method, comprising:
employing at least one processor configured to execute computer-executable instructions stored in a memory to perform the following acts:
constructing a hierarchical model structure automatically as a function of data accumulation over time, the structure includes a set of predictive models across a plurality of levels of granularity.

### Claim 2 (depends on 1)

The method of claim 1 further comprises identifying a candidate set of models in the structure for a particular purpose.

### Claim 3 (depends on 2)

The method of claim 2 further comprises:
comparing performance of the models in the candidate set of models; and
identifying one of the models as a best model based on performance.

### Claim 4 (depends on 3)

The method of claim 3 further comprises activating the best model for use.

### Claim 5 (depends on 4)

The method of claim 4 further comprising predicting a probability of an outcome based on the best model.

### Claim 6 (depends on 5)

The method of claim 5, predicting a probability that a transaction is fraudulent.

### Claim 7 (depends on 1)

The method of claim 1 further comprises acquiring attribute data related to a transaction.

### Claim 8 (depends on 7)

The method of claim 7 further comprises deriving additional attribute data from the acquired attribute data.

### Claim 9

A system, comprising:
a processor coupled to a memory, the processor configured to execute the following computer-executable components stored in the memory:
a first component configured to automatically predict an outcome based on an input and a predictive model; and
a second component configured to automatically update the model with a best performing model from a set of candidate models distributed across a plurality of hierarchical levels.

### Claim 10 (depends on 9)

The system of claim 9 further comprising a third component configured to automatically train the candidate models based on accumulated data.

### Claim 11 (depends on 10)

The system of claim 10, the accumulated data comprises input attributes and attributes derived from the input attributes.

### Claim 12 (depends on 9)

The system of claim 9, the first component is configured to predict a risk of fraud for an input transaction.

### Claim 13 (depends on 12)

The system of claim 12, the input transaction is a purchase.

### Claim 14 (depends on 9)

The system of claim 9, the first component is configured as a web-based service.

### Claim 15

A computer-readable storage medium having instructions stored thereon that enable at least one processor to perform a method upon execution of the instructions, the method comprising:
determining a set of candidate models for an input transaction from a set of predictive models segmented across two or more hierarchical levels;
computing performance metrics for the candidate models; and
identifying a top performing model from the set of candidate models based on a comparison of the performance metrics.

### Claim 16 (depends on 15)

The method of claim 15 further comprising activating the top performing model for use.

### Claim 17 (depends on 16)

The method of claim 16 further comprises predicting likelihood of an outcome based on real-time input and the top performing model.

### Claim 18 (depends on 15)

The method of claim 15 further comprising accumulating data associated with input transactions.

### Claim 19 (depends on 18)

The method of claim 18 further comprising re-training the set of predictive models automatically based on the accumulated data.

### Claim 20 (depends on 18)

The method of claim 18 further comprises adding a new predictive model to a hierarchical level as a function of accumulated data.