# US 2024/0202281 A1 · Drop-layer based vector augmentation

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

- Number: US 2024/0202281 A1
- Document ID: US20240202281A1
- Status: Published application
- Kind: A1
- Application number: 18/067,116
- Filed: 2022-12-16
- Published: 2024-06-20
- Earliest claimed: 2022-12-16
- Assignees: Sap SE
- Applicants: Sap SE
- Inventors: Tassilo Klein; Moin Nabi
- Web page: https://patentgrep.com/patent/US20240202281A1

## CPC

- 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
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A method comprises generating vectors of sentences, the sentences including text, performing an augmentation operation on a first vector of the vectors and a second vector of the vectors, the augmentation operation comprising: merging the first vector with the second vector such that at least a first element of the first vector is combined with a second element of the second vector, wherein the first element is representative of a first sentence of the plurality of sentences and the second element represents a second sentence of the plurality of sentences, inputting, the first vector that is merged with the second vector, into a first layer, and generating a third vector that is based on the first vector that is merged with the second vector, the third vector including a third element that represents the first sentence and the second sentence.

## Claims

### Claim 1

A computer-implemented method comprising:
generating vectors comprising a plurality of elements representative of a plurality of sentences, the plurality of sentences including text that is included as part of training data;
performing an augmentation operation on a first vector of the vectors and a second vector of the vectors, the augmentation operation comprising:
merging, in accordance with a weighting factor, the first vector with the second vector such that at least a first element of the first vector is combined with a second element of the second vector, wherein the first element is representative of a first sentence of the plurality of sentences and the second element represents a second sentence of the plurality of sentences,
inputting, the first vector that is merged with the second vector, into a first layer; and
generating, using the first layer, a third vector that is based on the first vector that is merged with the second vector, the third vector including a third element that represents the first sentence and the second sentence.

### Claim 2 (depends on 1)

The computer-implemented method of claim 1, wherein the merging of the first vector with the second vector includes concatenating the first element representative of the first sentence with the second element.

### Claim 3 (depends on 2)

The computer-implemented method of claim 2, wherein the first element comprises a first plurality of numbers representative of one or more letters of the first sentence and the second element comprises a second plurality of numbers representative of one or more letters of the second sentence.

### Claim 4 (depends on 3)

The computer-implemented method of claim 3, wherein the third element includes a third plurality of numbers that comprises the first plurality of numbers and the second plurality of numbers.

### Claim 5 (depends on 1)

The computer-implemented method of claim 1, further comprising determining the weighting factor randomly.

### Claim 6 (depends on 5)

The computer-implemented method of claim 5, wherein the weighting factor has a value in a range between 0 and 1.

### Claim 7 (depends on 1)

The computer-implemented method of claim 1, further comprising:
generating additional vectors comprising a plurality of additional elements representative of a plurality of images included as part of the training data;
performing an additional augmentation operation on a fourth vector of the additional vectors and a fifth vector of the additional vectors, the augmentation operation comprising:
merging, in accordance with an additional weighting factor, the fourth vector with the fifth vector such that a fourth element of the fourth vector is combined with a fifth element of the fifth vector, wherein the fourth element represents a first characteristic of an image of the plurality of images and the fifth element represents a second characteristic of an additional image,
inputting, the fourth vector that is merged with the fifth vector, into a second layer; and
generating, using the second layer, a sixth vector that is based on the fourth vector that is merged with the fifth vector, the sixth vector including a sixth element that represents the first characteristic of the image and the second characteristic of the additional image.

### Claim 8 (depends on 7)

The method of claim 7, wherein the weighting factor has a value in a range between 0 and 1.

### Claim 9 (depends on 7)

The method of claim 7, wherein:
the first characteristic corresponds to a color of the image and the second characteristic corresponds to an additional color in the additional image, and
wherein the fourth element comprises numbers representing the color and the fifth element comprises additional numbers representing the additional color.

### Claim 10 (depends on 9)

The method of claim 9, wherein the sixth element includes the first characteristic and the second characteristic, the sixth element comprises the numbers representing the color and the additional numbers representing the additional color.

### Claim 11

A system comprising:
at least one data processor; and
at least one memory storing instructions, which when executed by the at least one data processor, cause operations comprising:
generating vectors comprising a plurality of elements representative of a plurality of sentences, the plurality of sentences including text that is included as part of training data;
performing an augmentation operation on a first vector of the vectors and a second vector of the vectors, the augmentation operation comprising:
merging, in accordance with a weighting factor, the first vector with the second vector such that at least a first element of the first vector is combined with a second element of the second vector,
wherein the first element is representative of a first sentence of the plurality of sentences and the second element represents a second sentence of the plurality of sentences,
inputting, the first vector that is merged with the second vector, into a first layer; and
generating, using the first layer, a third vector that is based on the first vector that is merged with the second vector, the third vector including a third element that represents the first sentence and the second sentence.

### Claim 12 (depends on 11)

The system of claim 11, wherein the merging of the first vector with the second vector includes concatenating the first element representative of the first sentence with the second element.

### Claim 13 (depends on 12)

The system of claim 12, wherein the first element comprises a first plurality of numbers representative of one or more letters of the first sentence and the second element comprises a second plurality of numbers representative of one or more letters of the second sentence.

### Claim 14 (depends on 13)

The system of claim 13, wherein the third element includes a third plurality of numbers that comprises the first plurality of numbers and the second plurality of numbers.

### Claim 15 (depends on 11)

The system of claim 11, wherein the weighting factor has a value in a range between 0 and 1.

### Claim 16 (depends on 11)

The system of claim 11, wherein the operations further comprise:
generating additional vectors comprising a plurality of additional elements representative of a plurality of images included as part of the training data;
performing an additional augmentation operation on a fourth vector of the additional vectors and a fifth vector of the additional vectors, the augmentation operation comprising:
merging, in accordance with an additional weighting factor, the fourth vector with the fifth vector such that a fourth element of the fourth vector is combined with a fifth element of the fifth vector,
wherein the fourth element represents a first characteristic of an image of the plurality of images and the fifth element represents a second characteristic of an additional image,
inputting, the fourth vector that is merged with the fifth vector, into a second layer; and
generating, using the second layer, a sixth vector that is based on the fourth vector that is merged with the fifth vector, the sixth vector including a sixth element that represents the first characteristic of the image and the second characteristic of the additional image.

### Claim 17 (depends on 16)

The system of claim 16, wherein the weighting factor has a value in a range between 0 and 1.

### Claim 18 (depends on 16)

The system of claim 16, wherein:
the first characteristic corresponds to a color of the image and the second characteristic corresponds to an additional color in the additional image, and
wherein the fourth element comprises numbers representing the color and the fifth element comprises additional numbers representing the additional color.

### Claim 19 (depends on 18)

The system of claim 18, wherein the sixth element includes the first characteristic and
the second characteristic, the sixth element comprises the numbers representing the color and the additional numbers representing the additional color.

### Claim 20

At least one non-transitory computer readable media storing instructions that, when executed by at least one processor, cause the at least one processor to perform operations comprising:
generating vectors comprising a plurality of elements representative of a plurality of sentences, the plurality of sentences including text that is included as part of training data;
performing an augmentation operation on a first vector of the vectors and a second vector of the vectors, the augmentation operation comprising:
merging, in accordance with a weighting factor, the first vector with the second vector such that at least a first element of the first vector is combined with a second element of the second vector,
wherein the first element is representative of a first sentence of the plurality of sentences and the second element represents a second sentence of the plurality of sentences,
inputting, the first vector that is merged with the second vector, into a first layer; and
generating, using the first layer, a third vector that is based on the first vector that is merged with the second vector, the third vector including a third element that represents the first sentence and the second sentence.
