# US 12,015,691 B2 · Security as a service for machine learning

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

- Number: US 12,015,691 B2
- Document ID: US12015691B2
- Status: Granted
- Kind: B2
- Application number: 17/483,051
- Filed: 2021-09-23
- Published: 2023-03-23
- Granted: 2024-06-18
- Earliest claimed: 2021-09-23
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Mansura Habiba; Gokhan Sagirlar; Martin Wistuba
- Web page: https://patentgrep.com/patent/US12015691B2

## CPC

- H04L 9/008: {involving homomorphic encryption}
- G06F 21/55: Detecting local intrusion or implementing counter-measures
- G06F 21/6245: {Protecting personal data, e.g. for financial or medical purposes}
- G06N 3/04: Architecture, e.g. interconnection topology
- G06N 3/08: Learning methods
- G06N 3/094: Adversarial learning
- H04L 9/0662: {with particular pseudorandom sequence generator}
- G06F 2221/032: Protect output to user by software means
- G06N 3/088: Non-supervised learning, e.g. competitive learning
- H04L 2209/08: Randomization, e.g. dummy operations or using noise
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Embodiments are disclosed for a method. The method includes validating training data that is provided for training a machine learning model using ordinary differential equations. The method further includes generating pre-processed training data from the validated training data by generating encrypted training data from the validated training data using homomorphic encryption and generating random noise based on the validated training data. The method also includes training the machine learning model adversarially with the pre-processed training data.

## Claims

### Claim 1

A system comprising:
one or more computer processing circuits; and
one or more computer-readable storage media storing program instructions which, when executed by the one or more computer processing circuits, are configured to cause the one or more computer processing circuits to perform a method comprising:
validating, before training a machine learning model using a deep learning network, training data that is provided for training a machine learning model using ordinary differential equations;
generating pre-processed training data from the validated training data by:
generating encrypted training data from the validated training data using homomorphic encryption; and
generating random noise based on the validated training data; and
training the machine learning model adversarially with the pre-processed training data.

### Claim 2 (depends on 1)

The system of claim 1, the method further comprising generating a plurality of adversarial examples from the pre-processed training data.

### Claim 3 (depends on 1)

The system of claim 1, the method further comprising providing the training as a cloud service.

### Claim 4 (depends on 1)

The system of claim 1, further comprising:
a data validation layer that validates the training data;
a pre-processing layer that generates the pre-processed training data; and
a data training layer that trains the machine learning model.

### Claim 5 (depends on 1)

The system of claim 1, the method further comprising:
receiving feedback for one or more predictions by the machine learning model based on a quality of service agreement for the machine learning model; and
updating the machine learning model based on the feedback.

### Claim 6 (depends on 1)

The system of claim 1, wherein validating the training data comprises generating a validation report that identifies potentially poisoned training data.

### Claim 7

A computer-implemented method, comprising:
validating, before training a machine learning model using a deep learning network, training data that is provided for training a machine learning model using ordinary differential equations;
generating pre-processed training data from the validated training data by:
generating encrypted training data from the validated training data using homomorphic encryption; and
generating random noise based on the validated training data; and
training the machine learning model adversarially with the pre-processed training data.

### Claim 8 (depends on 7)

The method of claim 7, further comprising generating a plurality of adversarial examples from the pre-processed training data.

### Claim 9 (depends on 7)

The method of claim 7, further comprising providing the training as a cloud service.

### Claim 10 (depends on 7)

The method of claim 7, wherein:
a data validation layer validates the training data;
a pre-processing layer generates the pre-processed training data; and
a data training layer trains the machine learning model.

### Claim 11 (depends on 7)

The method of claim 7, further comprising:
receiving feedback for one or more predictions by the machine learning model based on a quality of service agreement for the machine learning model; and
updating the machine learning model based on the feedback.

### Claim 12 (depends on 7)

The method of claim 7, wherein validating the training data comprises generating a validation report that identifies potentially poisoned training data.

### Claim 13

A computer program product comprising one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instructions comprising instructions configured to cause one or more processors to perform a method comprising:
validating, before training a machine learning model using a deep learning network, training data that is provided for training a machine learning model using ordinary differential equations;
generating pre-processed training data from the validated training data by:
generating encrypted training data from the validated training data using homomorphic encryption; and
generating random noise based on the validated training data;
generating a plurality of adversarial examples from the pre-processed training data; and
training the machine learning model adversarially with the pre-processed training data and the plurality of adversarial examples.

### Claim 14 (depends on 13)

The computer program product of claim 13, the method further comprising providing the training as a cloud service.

### Claim 15 (depends on 13)

The computer program product of claim 13, wherein:
a data validation layer validates the training data;
a pre-processing layer generates the pre-processed training data; and
a data training layer trains the machine learning model.

### Claim 16 (depends on 13)

The computer program product of claim 13, the method further comprising:
receiving feedback for one or more predictions by the machine learning model based on a quality of service agreement for the machine learning model; and
updating the machine learning model based on the feedback.

### Claim 17 (depends on 13)

The computer program product of claim 13, wherein validating the training data comprises generating a validation report that identifies potentially poisoned training data.