# US 2026/0087345 A1 · Model parameter optimization method, related system, and storage medium

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

- Number: US 2026/0087345 A1
- Document ID: US20260087345A1
- Status: Published application
- Kind: A1
- Application number: 19/408,496
- Filed: 2025-12-04
- Published: 2026-03-26
- Earliest claimed: 2023-06-05
- Assignees: Huawei Technologies Co., Ltd.
- Applicants: Huawei Technologies Co., Ltd.
- Inventors: Ziyu Wang; Jiaxun Lu; Bowen Ding; Daqi Xie; Yunfeng Shao; Qibin Wu
- Web page: https://patentgrep.com/patent/US20260087345A1

## CPC

- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06N 3/096: Transfer learning
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/00: Machine learning
- G06F 30/27: using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Embodiments of this application provide a model parameter optimization method, a related system, and a storage medium, to avoid introducing a complex hyperparameter for balancing an amount of prior data and an amount of posterior data during parameter optimization, and improve model accuracy.

## Claims

### Claim 1

A model parameter optimization method, comprising:
obtaining a plurality of reference values of a parameter of a first model based on a prior data pair of the first model and a posterior data pair of the first model;
traversing the plurality of reference values based on preset model evaluation information, the first model, and the prior data pair, to obtain at least one target value of the parameter, wherein a loss value that is of a model on the prior data pair and that corresponds to the target value is not greater than an upper bound of a loss value of a model on the prior data pair in the preset model evaluation information, and/or a test accuracy value that is of the model on the prior data pair and that corresponds to the target value is not less than a lower bound of a test accuracy value of the model on the prior data pair in the preset model evaluation information;
determining an optimized value of the parameter based on the at least one target value, the prior data pair, and the posterior data pair; and
updating the parameter of the first model based on the optimized value.

### Claim 2 (depends on 1)

The method according to claim 1, wherein determining the optimized value of the parameter based on the at least one target value, the prior data pair, and the posterior data pair comprises:
obtaining a prior distribution of the parameter of the first model based on the prior data pair;
obtaining a loss value of the first model on the posterior data pair based on the posterior data pair and the at least one target value;
obtaining a posterior distribution of the parameter based on the prior distribution of the parameter and the loss value of the first model on the posterior data pair; and
determining the optimized value of the parameter based on the posterior distribution of the parameter.

### Claim 3 (depends on 2)

The method according to claim 2, wherein a value that is of the parameter and that corresponds to a maximum probability in the posterior distribution of the parameter is determined as the optimized value of the parameter.

### Claim 4 (depends on 2)

The method according to claim 2, wherein determining the optimized value of the parameter based on the posterior distribution of the parameter comprises:
performing sampling based on the posterior distribution of the parameter to obtain a plurality of values of the parameter; and
performing weighted averaging processing on the plurality of values of the parameter, to obtain the optimized value of the parameter.

### Claim 5 (depends on 1)

The method according to claim 1, wherein the parameter is all parameters of the first model.

### Claim 6 (depends on 1)

The method according to claim 1, wherein the parameter is a part of parameters of the first model, the first model sequentially comprises a feature extraction model and a classification model, and the part of parameters are parameters corresponding to the classification model;
traversing the plurality of reference values based on the preset model evaluation information, the first model, and the prior data pair, to obtain the at least one target value of the parameter comprises:
obtaining a prior feature data pair based on the prior data pair and the feature extraction model; and
traversing the plurality of reference values based on the preset model evaluation information, the classification model, and the prior feature data pair, to obtain the at least one target value of the parameter; and
determining the optimized value of the parameter based on the at least one target value, the prior data pair, and the posterior data pair comprises:
obtaining a posterior feature data pair based on the posterior data pair and the feature extraction model; and
determining the optimized value of the parameter based on the at least one target value, the prior feature data pair, the posterior feature data pair, and the classification model.

### Claim 7

A model parameter optimization apparatus, comprising a processor and a memory, wherein the memory is configured to store program code, and the program code, when executed by the processor, causes the processor to:
obtain a plurality of reference values of a parameter of a first model based on a prior data pair of the first model and a posterior data pair of the first model;
traverse the plurality of reference values based on preset model evaluation information, the first model, and the prior data pair, to obtain at least one target value of the parameter, wherein a loss value that is of a model on the prior data pair and that corresponds to the target value is not greater than an upper bound of a loss value of a model on the prior data pair in the preset model evaluation information, and/or a test accuracy value that is of the model on the prior data pair and that corresponds to the target value is not less than a lower bound of a test accuracy value of the model on the prior data pair in the preset model evaluation information;
determine an optimized value of the parameter based on the at least one target value, the prior data pair, and the posterior data pair; and
update the parameter of the first model based on the optimized value.

### Claim 8 (depends on 7)

The apparatus according to claim 7, wherein determining the optimized value of the parameter based on the at least one target value, the prior data pair, and the posterior data pair comprises:
obtaining a prior distribution of the parameter of the first model based on the prior data pair;
obtaining a loss value of the first model on the posterior data pair based on the posterior data pair and the at least one target value;
obtaining a posterior distribution of the parameter based on the prior distribution of the parameter and the loss value of the first model on the posterior data pair; and
determining the optimized value of the parameter based on the posterior distribution of the parameter.

### Claim 9 (depends on 8)

The apparatus according to claim 8, wherein a value that is of the parameter and that corresponds to a maximum probability in the posterior distribution of the parameter is determined as the optimized value of the parameter.

### Claim 10 (depends on 8)

The apparatus according to claim 8, wherein determining the optimized value of the parameter based on the posterior distribution of the parameter comprises:
performing sampling based on the posterior distribution of the parameter to obtain a plurality of values of the parameter; and
performing weighted averaging processing on the plurality of values of the parameter, to obtain the optimized value of the parameter.

### Claim 11 (depends on 7)

The apparatus according to claim 7, wherein the parameter is all parameters of the first model.

### Claim 12 (depends on 7)

The apparatus according to claim 7, wherein the parameter is a part of parameters of the first model, the first model sequentially comprises a feature extraction model and a classification model, and the part of parameters are parameters corresponding to the classification model;
traversing the plurality of reference values based on the preset model evaluation information, the first model, and the prior data pair, to obtain the at least one target value of the parameter comprises:
obtaining a prior feature data pair based on the prior data pair and the feature extraction model; and
traversing the plurality of reference values based on the preset model evaluation information, the classification model, and the prior feature data pair, to obtain the at least one target value of the parameter; and
determining the optimized value of the parameter based on the at least one target value, the prior data pair, and the posterior data pair comprises:
obtaining a posterior feature data pair based on the posterior data pair and the feature extraction model; and
determining the optimized value of the parameter based on the at least one target value, the prior feature data pair, the posterior feature data pair, and the classification model.

### Claim 13

A computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program, when executed by a processor, causes the processor to:
obtain a plurality of reference values of a parameter of a first model based on a prior data pair of the first model and a posterior data pair of the first model;
traverse the plurality of reference values based on preset model evaluation information, the first model, and the prior data pair, to obtain at least one target value of the parameter, wherein a loss value that is of a model on the prior data pair and that corresponds to the target value is not greater than an upper bound of a loss value of a model on the prior data pair in the preset model evaluation information, and/or a test accuracy value that is of the model on the prior data pair and that corresponds to the target value is not less than a lower bound of a test accuracy value of the model on the prior data pair in the preset model evaluation information;
determine an optimized value of the parameter based on the at least one target value, the prior data pair, and the posterior data pair; and
update the parameter of the first model based on the optimized value.

### Claim 14 (depends on 13)

The storage medium according to claim 13, wherein determining the optimized value of the parameter based on the at least one target value, the prior data pair, and the posterior data pair comprises:
obtaining a prior distribution of the parameter of the first model based on the prior data pair;
obtaining a loss value of the first model on the posterior data pair based on the posterior data pair and the at least one target value;
obtaining a posterior distribution of the parameter based on the prior distribution of the parameter and the loss value of the first model on the posterior data pair; and
determining the optimized value of the parameter based on the posterior distribution of the parameter.

### Claim 15 (depends on 14)

The storage medium according to claim 14, wherein a value that is of the parameter and that corresponds to a maximum probability in the posterior distribution of the parameter is determined as the optimized value of the parameter.

### Claim 16 (depends on 14)

The storage medium according to claim 14, wherein determining the optimized value of the parameter based on the posterior distribution of the parameter comprises:
performing sampling based on the posterior distribution of the parameter to obtain a plurality of values of the parameter; and
performing weighted averaging processing on the plurality of values of the parameter, to obtain the optimized value of the parameter.

### Claim 17 (depends on 13)

The storage medium according to claim 13, wherein the parameter is all parameters of the first model.

### Claim 18 (depends on 13)

The storage medium according to claim 13, wherein the parameter is a part of parameters of the first model, the first model sequentially comprises a feature extraction model and a classification model, and the part of parameters are parameters corresponding to the classification model;
traversing the plurality of reference values based on the preset model evaluation information, the first model, and the prior data pair, to obtain the at least one target value of the parameter comprises:
obtaining a prior feature data pair based on the prior data pair and the feature extraction model; and
traversing the plurality of reference values based on the preset model evaluation information, the classification model, and the prior feature data pair, to obtain the at least one target value of the parameter; and
determining the optimized value of the parameter based on the at least one target value, the prior data pair, and the posterior data pair comprises:
obtaining a posterior feature data pair based on the posterior data pair and the feature extraction model; and
determining the optimized value of the parameter based on the at least one target value, the prior feature data pair, the posterior feature data pair, and the classification model.
