# US 2024/0005341 A1 · Customer experience perception based on federated learning

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

- Number: US 2024/0005341 A1
- Document ID: US20240005341A1
- Status: Published application
- Kind: A1
- Application number: 18/030,949
- Filed: 2021-05-08
- Published: 2024-01-04
- Earliest claimed: 2021-05-08
- Assignees: Asiainfo Technologies (China), Inc.
- Applicants: Asiainfo Technologies (China), Inc.
- Inventors: Ye OUYANG; Aidong YANG; Xiangde MENG; Mohan WU
- Web page: https://patentgrep.com/patent/US20240005341A1

## CPC

- G06Q 30/0201: Market modelling; Market analysis; Collecting market data
- G06N 20/00: Machine learning
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

The present disclosure relates to customer experience perception based on federated learning. An electronic device for a service provider in a wireless communication system, the service provider capable of implementing customer experience perception with at least one other service provider in the wireless communication system through federated learning, the electronic device comprises a processing circuit configured to: determine training data for training a global model related to the customer experience perception based on the federated learning, wherein the training data is related to the customer's experience for a specific service/product/business; perform local model training by using the determined training data to obtain intermediate information related to local model training; and transmit the intermediate information to a coordinator, so that the coordinator can train a customer experience perception model by aggregating the intermediate information from the provider and at least one other provider.

## Claims

### Claim 1

An electronic device for a service provider in a wireless communication system, the service provider capable of implementing customer experience perception with at least one other service provider in the wireless communication system through federated learning, the electronic device comprises a processing circuit configured to:
determine training data for training a global model related to the customer experience perception based on the federated learning, wherein the training data is related to the customer's experience for a specific service/product/business;
perform local model training by using the determined training data to obtain intermediate information related to local model training; and
transmit the intermediate information to a coordinator, so that the coordinator can train a customer experience perception model by aggregating the intermediate information from the provider and at least one other provider.

### Claim 2 (depends on 1)

The electronic device of claim 1, wherein the service provider and the at least one other service provider are selected based on information about data overlapping status thereamong.

### Claim 3 (depends on 2)

The electronic device of claim 2, wherein information about data overlapping status may include at least one of data ID overlapping ratio and data feature overlapping ratio, and
wherein two service providers in the system, when at least one of data ID overlapping ratio and data feature overlapping ratio is larger than a corresponding threshold, are suitable for participating in the customer experience perception based on the federated learning.

### Claim 4 (depends on 1)

The electronic device of claim 1, wherein the type of federated learning or federated learning algorithm is determined based on the information about data overlapping status among the service provider and the at least one other service provider.

### Claim 5 (depends on 3)

The electronic device of claim 3, wherein information about data overlapping status may include at least one of data ID overlapping ratio and data feature overlapping ratio, and
wherein when the data ID overlapping ratio is larger while data feature overlapping ratio is smaller, a vertical federated learning is utilized, and when the data ID overlapping ratio is smaller while data feature overlapping ratio is larger, a horizontal federated learning is utilized.

### Claim 6 (depends on 1)

The electronic device of claim 1, wherein the training data is data at the service provider corresponding to common identifiers which are matched identifiers between the local data of the service provider and the local data of the at least one other service provider.

### Claim 7 (depends on 1)

The electronic device of claim 1, wherein the processing circuit is further configured to:
acquire information about identifiers of samples of local data from the at least one other service provider,
determine identifier intersection between identifiers of samples of local data of the service provider and identifiers of samples of local data from the at least one other service provider as the common identifiers, and
transmit relevant information about the common identifiers to the at least one other service provider.

### Claim 8 (depends on 1)

The electronic device of claim 1, wherein the processing circuit is further configured to:
transmit information about identifiers of samples of local data of the service provider to an entity in the system for determining the common identifiers; and
receive relevant information about the common identifiers from the entity.

### Claim 9 (depends on 1)

The electronic device of claim 1, wherein the intermediate information includes information about the result of local model training based on the customer experience related sample data at the service provider corresponding to common identifiers, and
wherein the result of local model training includes at least one of model parameters and model performance loss.

### Claim 10 (depends on 1)

The electronic device of claim 1, wherein the processing circuit is further configured to:
receive information about the trained model from the coordinator, and
update the local model at the service provider based on the information about the trained model.

### Claim 11 (depends on 1)

The electronic device of claim 1, wherein the local model training at the service provider and the customer experience perception model training at the coordinator are iteratively implemented, until a specific termination condition is satisfied.

### Claim 12 (depends on 11)

The electronic device of claim 11, wherein the specific termination condition is one of:
whether the iteration number reach a specific number; and
whether the model performance loss is smaller than a threshold.

### Claim 13 (depends on 1)

The electronic device of claim 1, wherein the processing circuit is further configured to:
perform local customer experience perception evaluation based on customer experience related data, and
transmit information about the local customer experience perception evaluation to the coordinator, so that the coordinator implements global customer experience perception evaluation by aggregating the information about the local customer experience perception evaluation from the service provider and the at least one other service provider.

### Claim 14 (depends on 1)

The electronic device of claim 1, wherein the local model at the service provider is a part of the global model to be trained or the global model.

### Claim 15 (depends on 1)

The electronic device of claim 1, wherein at least one of the information about identifiers, the relevant information about the common identifiers, the intermediate information related to local model training, and the information about the local customer experience perception evaluation are encrypted.

### Claim at position 16

16-22. (canceled)

### Claim 23

A method for a service provider in a wireless communication system, the service provider capable of implementing customer experience perception with at least one other service provider in the wireless communication system through federated learning, the method comprises:
determining training data for training a global model related to the customer experience perception based on the federated learning, wherein the training data is related to the customer's experience for a specific service/product/business;
performing local model training by using the determined training data to obtain intermediate information related to local model training; and
transmitting the intermediate information to a coordinator, so that the coordinator can train a customer experience perception model by aggregating the intermediate information from the provider and at least one other provider.

### Claim 24

(canceled)

### Claim 25 (depends on 23)

A device comprising:
one or more processors; and
one or more storage media storing instructions that, when executed by the one or more processors, cause the method according to claim 23 to be performed.

### Claim at position 20

26-27. (canceled)

### Claim 28 (depends on 23)

The method of claim 23, wherein the method further comprises:
acquiring information about identifiers of samples of local data from the at least one other service provider,
determining identifier intersection between identifiers of samples of local data of the service provider and identifiers of samples of local data from the at least one other service provider as the common identifiers, and
transmitting relevant information about the common identifiers to the at least one other service provider.

### Claim 29 (depends on 23)

The method of claim 23, wherein the method further comprises:
receiving information about the trained model from the coordinator, and
updating the local model at the service provider based on the information about the trained model.

### Claim 30 (depends on 28)

A device comprising:
one or more processors; and
one or more storage media storing instructions that, when executed by the one or more processors, cause the method according to claim 28 to be performed.
