US 12,230,012 B2
Machine learning system and method, integration server, information processing apparatus, program, and inference model creation method
Daiki Uehara, Tokyo (JP)
Assigned to FUJIFILM Corporation, Tokyo (JP)
Filed by FUJIFILM Corporation, Tokyo (JP)
Filed on Apr. 17, 2022, as Appl. No. 17/722,383.
Application 17/722,383 is a continuation of application No. PCT/JP2020/038694, filed on Oct. 14, 2020.
Claims priority of application No. 2019-192548 (JP), filed on Oct. 23, 2019.
Prior Publication US 2022/0237898 A1, Jul. 28, 2022
Int. Cl. G06V 10/774 (2022.01); G06N 3/044 (2023.01); G06N 3/045 (2023.01); G06N 3/063 (2023.01); G06N 3/08 (2023.01); G06N 5/022 (2023.01); G06V 10/764 (2022.01); G06V 10/776 (2022.01); G06V 10/82 (2022.01); G16H 30/40 (2018.01); G16H 50/20 (2018.01)
CPC G06V 10/774 (2022.01) [G06N 5/022 (2013.01); G06V 10/764 (2022.01); G06V 10/776 (2022.01); G06V 10/82 (2022.01); G06V 2201/03 (2022.01)] 24 Claims
OG exemplary drawing
 
1. A machine learning system comprising:
a plurality of client terminals; and
an integration server,
wherein each of the plurality of client terminals includes
a terminal-side processor configured to:
classify data stored in a data storage apparatus of a medical institution based on an acquisition condition of the data to classify learning data into each data group acquired under the same or a similar acquisition condition, the acquisition condition including condition concerning apparatus used to generate the data;
execute machine learning of a learning model for each learning data group classified into each condition category of the same or a similar acquisition condition; and
transmit learning results of the learning model executed for each learning data group and condition information regarding the acquisition condition of the learning data group used for the learning, to the integration server, and
the integration server includes
a trained master model, and
a server-side processor configured to:
synchronize the learning model of each client terminal side with the master model before each of the plurality of client terminals trains the learning model;
receive the learning results of the learning model and the condition information from each of the plurality of client terminals;
classify the learning results into each condition category;
integrate the learning results for each condition category to create a plurality of master model candidates; and
evaluate an inference accuracy of each of the plurality of master model candidates,
wherein the data includes inspection data acquired by using an inspection apparatus, and the acquisition condition includes an inspection condition under which the inspection data is acquired,
wherein the inspection condition includes condition concerning the inspection apparatus used for inspection.