# US 2023/0057593 A1 · Compensating for vulnerabilities in machine learning algorithms

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

- Number: US 2023/0057593 A1
- Document ID: US20230057593A1
- Status: Published application
- Kind: A1
- Application number: 17/408,385
- Filed: 2021-08-21
- Published: 2023-02-23
- Earliest claimed: 2021-08-21
- Assignees: AT&T Intellectual Property I, L.P.
- Applicants: AT&T Intellectual Property I, L.P.
- Inventors: Balachander Krishnamurthy; Subhabrata Majumdar
- Web page: https://patentgrep.com/patent/US20230057593A1

## CPC

- G06N 20/00: Machine learning
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A method performed by a processing system including at least one processor includes obtaining an output of a machine learning algorithm, identifying a vulnerability in the output of the machine learning algorithm, wherein the vulnerability relates to a bias in the output, integrating auxiliary data from an auxiliary data source of a plurality of auxiliary data sources into the machine learning algorithm to try to compensate for the vulnerability, determining whether the integrating has compensated for the vulnerability, and generating a runtime output using the machine learning algorithm when the processing system determines that the integrating has compensated for the vulnerability.

## Claims

### Claim 1

A method comprising:
obtaining, by a processing system including at least one processor, an output of a machine learning algorithm;
identifying, by the processing system, a vulnerability in the output of the machine learning algorithm, wherein the vulnerability relates to a bias in the output;
integrating, by the processing system, auxiliary data from an auxiliary data source of a plurality of auxiliary data sources into the machine learning algorithm to try to compensate for the vulnerability;
determining, by the processing system, whether the integrating has compensated for the vulnerability; and
generating, by the processing system, a runtime output using the machine learning algorithm when the processing system determines that the integrating has compensated for the vulnerability.

### Claim 2 (depends on 1)

The method of claim 1, wherein the output comprises at least one of: generated multimedia content, a list of prioritized samples, or attributes and values that are considered high-value by a machine learning model or based on a domain knowledge.

### Claim 3 (depends on 1)

The method of claim 1, wherein the bias is a result of a misrepresentation of at least one aspect of a sample provided to the machine learning algorithm.

### Claim 4 (depends on 3)

The method of claim 3, wherein the misrepresentation is due to at least one of:
a bias of a human who labeled the sample, a bias in a sampling process used to generate the sample, or a systemic bias.

### Claim 5 (depends on 1)

The method of claim 1, wherein the integrating comprises retraining the machine learning algorithm using the auxiliary data as training data.

### Claim 6 (depends on 5)

The method of claim 5, wherein the integrating further comprises weighting the auxiliary data more heavily than training data originally used to train the machine learning algorithm during the retraining.

### Claim 7 (depends on 5)

The method of claim 5, wherein the auxiliary data source comprises a collection of ground-truth data that is used to enhance a degree of confidence in the output of the machine learning algorithm.

### Claim 8 (depends on 7)

The method of claim 7, wherein the auxiliary data source is curated specifically to target the vulnerability.

### Claim 9 (depends on 8)

The method of claim 8, wherein each auxiliary data source of the plurality of auxiliary data sources is curated to target a different potential vulnerability of a plurality of potential vulnerabilities.

### Claim 10 (depends on 1)

The method of claim 1, wherein the integrating comprises utilizing the auxiliary data as an alarm mechanism to detect when the output of the machine learning algorithm differs by more than a threshold from items of the auxiliary data that have input features similar to the output of the machine learning algorithm.

### Claim 11 (depends on 1)

The method of claim 1, wherein the determining comprises re-running the machine learning algorithm and checking an output of the re-running for the vulnerability.

### Claim 12 (depends on 1)

The method of claim 1, further comprising:
repeating, by the processing system subsequent to the determining but prior to the generating, the integrating when the processing system determines that the integrating has not compensated for the vulnerability.

### Claim 13 (depends on 1)

The method of claim 1, wherein the vulnerability is a previously unknown vulnerability, and the identifying comprises receiving a signal from a human domain expert who has detected the vulnerability.

### Claim 14 (depends on 13)

The method of claim 13, wherein the vulnerability is added to a list of known potential vulnerabilities against which a future machine learning algorithm is checked.

### Claim 15 (depends on 1)

The method of claim 1, wherein the auxiliary data source comprises a newly discovered auxiliary data source, and wherein the auxiliary data is stored for use in compensating for a vulnerability in a future machine learning algorithm.

### Claim 16 (depends on 1)

The method of claim 1, further comprising, subsequent to the identifying but prior to the integrating:
prioritizing, by the processing system, the vulnerability relative to a plurality of vulnerabilities that were identified in the output of the machine learning algorithm, wherein the prioritizing indicates which vulnerabilities of the plurality of vulnerabilities are most important to compensate for.

### Claim 17 (depends on 16)

The method of claim 16, wherein an importance of a respective vulnerability of the plurality of vulnerabilities is defined in terms of an impact on a performance of the machine learning algorithm.

### Claim 18 (depends on 16)

The method of claim 16, wherein an importance of a respective vulnerability of the plurality of vulnerabilities is defined in terms of a fairness consideration.

### Claim 19

A non-transitory computer-readable medium storing instructions which, when executed by a processing system including at least one processor, cause the processing system to perform operations, the operations comprising:
obtaining an output of a machine learning algorithm;
identifying a vulnerability in the output of the machine learning algorithm, wherein the vulnerability relates to a bias in the output;
integrating auxiliary data from an auxiliary data source of a plurality of auxiliary data sources into the machine learning algorithm to try to compensate for the vulnerability;
determining whether the integrating has compensated for the vulnerability; and
generating a runtime output using the machine learning algorithm when the processing system determines that the integrating has compensated for the vulnerability.

### Claim 20

A device comprising:
a processing system including at least one processor; and
a non-transitory computer-readable medium storing instructions which, when executed by the processing system, cause the processing system to perform operations, the operations comprising:
obtaining an output of a machine learning algorithm;
identifying a vulnerability in the output of the machine learning algorithm, wherein the vulnerability relates to a bias in the output;
integrating auxiliary data from an auxiliary data source of a plurality of auxiliary data sources into the machine learning algorithm to try to compensate for the vulnerability;
determining whether the integrating has compensated for the vulnerability; and
generating a runtime output using the machine learning algorithm when the processing system determines that the integrating has compensated for the vulnerability.
