# US 2026/0093957 A1 · Generating semi-structured tabular data using llm-based variational autoencoder

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

- Number: US 2026/0093957 A1
- Document ID: US20260093957A1
- Status: Published application
- Kind: A1
- Application number: 18/903,052
- Filed: 2024-10-01
- Published: 2026-04-02
- Earliest claimed: 2024-10-01
- Assignees: Sap SE
- Applicants: Sap SE
- Inventors: Paul Mattes; Rajesh Vellore Arumugam; Yi Quan Zhou
- Web page: https://patentgrep.com/patent/US20260093957A1

## CPC

- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06F 16/2282: {Tablespace storage structures; Management thereof}
- G06N 3/045: Combinations of networks
- G06N 3/047: Probabilistic or stochastic networks
- G06N 3/088: Non-supervised learning, e.g. competitive learning
- G06N 3/094: Adversarial learning
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Methods, systems, and computer-readable storage media for during a fine-tuning phase, receiving a semi-structured data object including a set of columns and a set of rows, each row representing a record, at least one column recording unstructured data, pre-processing rows of the semi-structured data object to generate a set of text strings, each text string representing a respective row of the semi-structured data object, and executing an adversarial training process using the set of texts strings to fine-tune parameters of one or more large LLMs of a MTV-GAN to provide a fine-tuned encoder and a fine-tuned decoder, during a synthetic data generation phase, processing a latent vector through the fine-tuned decoder of a VAE of the MTV-GAN to generate at least a portion of the synthetic data, and, during a training phase, executing a training process to train the ML model using the synthetic data.

## Claims

### Claim 1

A computer-implemented method for training a machine learning (ML) model using synthetic data, the method being executed by one or more processors and comprising:
during a fine-tuning phase:
receiving a first semi-structured data object comprising a set of columns and a set of rows, each row representing a record, at least one column recording unstructured data;
pre-processing rows of the first semi-structured data object to generate a first set of text strings, each text string representing a respective row of the first semi-structured data object, and
executing an adversarial training process using the first set of texts strings to fine-tune parameters of one or more large language models (LLMs) of a multi-modal tabular data variable autoencoder (VAE) generative adversarial network (GAN) (MTV-GAN) to provide a fine-tuned encoder and a fine-tuned decoder;
during a synthetic data generation phase:
processing a latent vector through the fine-tuned decoder of a VAE of the MTV-GAN to generate at least a portion of the synthetic data; and
during a training phase:
executing a training process to train the ML model using the synthetic data.

### Claim 2 (depends on 1)

The method of claim 1, wherein the fine-tuned decoder decodes the latent vector to generate the at least a portion of the synthetic data.

### Claim 3 (depends on 1)

The method of claim 1, wherein the fine-tuned encoder encodes a text string to provide the latent vector.

### Claim 4 (depends on 1)

The method of claim 1, further comprising:
receiving a second semi-structured data object;
pre-processing rows of the second semi-structured data object to generate a second set of text strings, each text string representing a respective row of the second semi-structured data object; and
generating the latent vector using a text string of the second set of text string.

### Claim 5 (depends on 4)

The method of claim 4, wherein the at least a portion of the synthetic data is generated by populating a partial text string with missing values that are generated by the fine-tuned decoder.

### Claim 6 (depends on 1)

The method of claim 1, wherein each of the LLMs is trained prior to execution of the adversarial training process.

### Claim 7 (depends on 1)

The method of claim 1, wherein data recorded in the first semi-structured data object comprises multi-modal data.

### Claim 8

A non-transitory computer-readable storage medium coupled to one or more processors and having instructions stored thereon which, when executed by the one or more processors, cause the one or more processors to perform operations for training a machine learning (ML) model using synthetic data, the operations comprising:
during a fine-tuning phase:
receiving a first semi-structured data object comprising a set of columns and a set of rows, each row representing a record, at least one column recording unstructured data;
pre-processing rows of the first semi-structured data object to generate a first set of text strings, each text string representing a respective row of the first semi-structured data object, and
executing an adversarial training process using the first set of texts strings to fine-tune parameters of one or more large language models (LLMs) of a multi-modal tabular data variable autoencoder (VAE) generative adversarial network (GAN) (MTV-GAN) to provide a fine-tuned encoder and a fine-tuned decoder;
during a synthetic data generation phase:
processing a latent vector through the fine-tuned decoder of a VAE of the MTV-GAN to generate at least a portion of the synthetic data; and
during a training phase:
executing a training process to train the ML model using the synthetic data.

### Claim 9 (depends on 8)

The non-transitory computer-readable storage medium of claim 8, wherein the fine-tuned decoder decodes the latent vector to generate the at least a portion of the synthetic data.

### Claim 10 (depends on 8)

The non-transitory computer-readable storage medium of claim 8, wherein the fine-tuned encoder encodes a text string to provide the latent vector.

### Claim 11 (depends on 8)

The non-transitory computer-readable storage medium of claim 8, wherein operations further comprise:
receiving a second semi-structured data object;
pre-processing rows of the second semi-structured data object to generate a second set of text strings, each text string representing a respective row of the second semi-structured data object; and
generating the latent vector using a text string of the second set of text string.

### Claim 12 (depends on 11)

The non-transitory computer-readable storage medium of claim 11, wherein the at least a portion of the synthetic data is generated by populating a partial text string with missing values that are generated by the fine-tuned decoder.

### Claim 13 (depends on 8)

The non-transitory computer-readable storage medium of claim 8, wherein each of the LLMs is trained prior to execution of the adversarial training process.

### Claim 14 (depends on 8)

The non-transitory computer-readable storage medium of claim 8, wherein data recorded in the first semi-structured data object comprises multi-modal data.

### Claim 15

A system, comprising:
a computing device; and
a computer-readable storage device coupled to the computing device and having instructions stored thereon which, when executed by the computing device, cause the computing device to perform operations for natural language explanations for training a machine learning (ML) model using synthetic data, the operations comprising:
during a fine-tuning phase:
receiving a first semi-structured data object comprising a set of columns and a set of rows, each row representing a record, at least one column recording unstructured data;
pre-processing rows of the first semi-structured data object to generate a first set of text strings, each text string representing a respective row of the first semi-structured data object, and
executing an adversarial training process using the first set of texts strings to fine-tune parameters of one or more large language models (LLMs) of a multi-modal tabular data variable autoencoder (VAE) generative adversarial network (GAN) (MTV-GAN) to provide a fine-tuned encoder and a fine-tuned decoder;
during a synthetic data generation phase:
processing a latent vector through the fine-tuned decoder of a VAE of the MTV-GAN to generate at least a portion of the synthetic data; and
during a training phase:
executing a training process to train the ML model using the synthetic data.

### Claim 16 (depends on 15)

The system of claim 15, wherein the fine-tuned decoder decodes the latent vector to generate the at least a portion of the synthetic data.

### Claim 17 (depends on 15)

The system of claim 15, wherein the fine-tuned encoder encodes a text string to provide the latent vector.

### Claim 18 (depends on 15)

The system of claim 15, wherein operations further comprise:
receiving a second semi-structured data object;
pre-processing rows of the second semi-structured data object to generate a second set of text strings, each text string representing a respective row of the second semi-structured data object; and
generating the latent vector using a text string of the second set of text string.

### Claim 19 (depends on 18)

The system of claim 18, wherein the at least a portion of the synthetic data is generated by populating a partial text string with missing values that are generated by the fine-tuned decoder.

### Claim 20 (depends on 15)

The system of claim 15, wherein each of the LLMs is trained prior to execution of the adversarial training process.
