Method and apparatus for federated learning of artificial intelligence model using non-IID data

Inventors: Seok Kyu Kang (Suwon-si, KR); Yong Hoon Kang (Suwon-si, KR); Jee Hyong Lee (Seoul, KR)

Assignee: RESEARCH & BUSINESS FOUNDATION SUNGKYUNKWAN UNIVERSITY

G06N20/00 G06F21/6245

View Patent ↗

Loading inventors, assignments & file history…

Monitor This Case

Get email alerts when status or documents change.

Order Certified Copies

Most orders are placed with the USPTO same day — all within 24 business hours.

Order via The Patent Place →

Pre-filled with this patent's details

Quick Facts

Patent No.: US 12664464
App. No.: 17/900,248
Granted: Jun 23, 2026
Kind: B2

External Links

USPTO Patent Center↗ Full Patent PDF↗

Abstract

The present disclosure relates to a method and an apparatus for federated learning of an artificial intelligence model. According to an exemplary embodiment of the present disclosure, a federated learning method of an artificial intelligence model includes: training a first local artificial intelligence model and a second local artificial intelligence model using data sets of a first client and a second client among the plurality of clients; calculating performance values for the first local artificial intelligence model and the second local artificial intelligence model by transmitting the first local artificial intelligence model to the second client and transmitting the second local artificial intelligence model to the first client; comparing the performance values to remove one of the first client and the second client; and training a global model using a client which is not removed.

Claims (27)

1 . A federated learning method of an artificial intelligence model, comprising:

training a first local artificial intelligence model and a second local artificial intelligence model using data sets of a first client and a second client among the plurality of clients;

calculating performance values for the first local artificial intelligence model and the second local artificial intelligence model by transmitting the first local artificial intelligence model to the second client and transmitting the second local artificial intelligence model to the first client,

wherein a data set of the second local intelligence model is a non-independent identically distributed (Non-IID) data set having a small volume and a plurality of labels local to the second client;

comparing the performance values to remove the second client; and

training a VGG-9 global model in an image classification task using the first client which is not removed,

wherein the training of a first local artificial intelligence model and a second local artificial intelligence model includes:

training the first local artificial intelligence model and the second local artificial intelligence model by a local epoch which exceeds a predetermined reference number of times.

2 . The federated learning method of an artificial intelligence model according to claim 1 , wherein the calculating of performance values includes:

calculating a performance value of the first local artificial intelligence model using a data set of the second client and calculating a performance value of the second local artificial intelligence model using a data set of the first client.

3 . The federated learning method of an artificial intelligence model according to claim 1 , wherein the removing includes:

removing the second client when the following Equation 1 is satisfied as a comparison result of the performance value:

P 1→2≥ P 1→1> P 2→1 <Equation 1>

here, P1→2 is a performance obtained by testing the first local artificial intelligence model with the data set of the second client, P1→1 is a performance obtained by training the first local artificial intelligence model with the data set of the first client, and P2→1 is a performance obtained by testing the second local artificial intelligence model with the data set of the first client.

4 . A federated learning apparatus of an artificial intelligence model, comprising:

one or more processors which execute an instruction,

wherein the one or more processors perform:

training a first local artificial intelligence model and a second local artificial intelligence model using data sets of a first client and a second client among the plurality of clients, by a local learning unit;

wherein a data set of the second local intelligence model is a non-independent identically distributed (Non-IID) data set having a small volume and a plurality of labels local to the second client;

comparing the performance values to remove the second client, by a client selecting unit; and

training a VGG-9 global model in an image classification task using the first client which is not removed, by a federated learning unit,

wherein the local learning unit trains the first local artificial intelligence model and the second local artificial intelligence model by a local epoch which exceeds a predetermined reference number of times.

5 . The federated learning apparatus of an artificial intelligence model according to claim 4 , wherein the performance value calculating unit calculates a performance value of the first local artificial intelligence model using a data set of the second client and calculates a performance value of the second local artificial intelligence model using a data set of the first client.

6 . The federated learning apparatus of an artificial intelligence model according to claim 4 , wherein the client selecting unit removes the second client when the following Equation 1 is satisfied as a comparison result of the performance value:

P 1→2≥ P 1→1> P 2→1 <Equation 1>