Method for obtaining training data by supplementing user input
Disclosed is a method for managing a local model and a global model on an AI platform, the method performed by one or more processors of a computing device according to an exemplary embodiment of the present disclosure. the method may include: obtaining a plurality of sample data; receiving a first user input for a first data set included in the plurality of sample data; receiving a second user input for a second data set excluding the first data set among the plurality of sample data; and obtaining a supplemented second data set by supplementing the second user input based on the first user input.
1 . A method for obtaining data for training a neural network model, the method performed by one or more processors of a computing device, the method comprising:
obtaining sample data;
receiving a first user input for a first data set included the sample data;
training a neural network model for predicting labeling information based on the first data set and the first user input;
receiving a second user input for a second data set excluding the first data set among the sample data; and
supplementing the second user input based on the first user input and obtaining a supplemented second data set based on the supplemented second user input,
wherein the training the neural network model for predicting labeling information based on the first data set and the first user input includes:
obtaining a first-second predicted user input of a second time point based on a first-first user input of a first time point by using the neural network model; and
training the neural network model based on a first-second user input of the second time point included in the first user input and the first-second predicted user input,
wherein the second time point is after the first time point.
2 . The method of claim 1 , wherein the first data set included in the sample data includes:
a data set sampled based on diversity for the sample data.
3 . The method of claim 1 , wherein the receiving the first user input for the first data set included in the sample data includes:
receiving the first user input related to a labeling operation for the first data set included in the sample data.
4 . The method of claim 3 , wherein the receiving the first user input related to the labeling operation for the first data set included in the sample data includes:
receiving the first-first user input corresponding to the first time point for the first data set; and
receiving the first-second user input corresponding to the second time point after the first time point for the first data set.
5 . The method of claim 1 , wherein the supplementing the second user input based on the first user input and obtaining a supplemented second data set based on the supplemented second user input includes:
obtaining a first predicted user input based on the second user input by using the trained neural network model; and
supplementing the second user input based on the first predicted user input and obtaining the supplemented second data set based on the supplemented second user input.
6 . The method of claim 1 , further comprising:
performing an evaluation on the supplemented second data set;
obtaining a third data set included in the sample data based on a result of the evaluation;
receiving a third user input for the third data set; and
obtaining a supplemented third data set based on the third user input and the third data set.
7 . The method of claim 6 , wherein performing the evaluation on the supplemented second data set includes:
measuring uncertainty for the supplemented second data set.
8 . The method of claim 7 , wherein obtaining the third data set included in the sample data based on the result of the evaluation includes:
obtaining the third data set among the sample data requiring additional information collection based on the measured uncertainty.
9 . A non-transitory computer-readable storage medium storing a computer program including instructions that cause, wherein the computer program causes one or more processors to perform operations for obtaining data for training a neural network model when the computer program is executed by the one or more processors, the operations comprising: an operation of obtaining sample data; an operation of receiving a first user input for a first data set included in the sample data; an operation of training a neural network model for predicting labeling information based on the first data set and the first user input; an operation of receiving a second user input for a second data set excluding the first data set among the sample data; and an operation of supplementing the second user input based on the first user input and obtaining a supplemented second data set based on the supplemented second user input, wherein the operation of training the neural network model for predicting labeling information based on the first data set and the first user input includes: an operation of obtaining a first-second predicted user input of a second time point based on a first-first user input of a first time point by using the neural network model; and an operation of training the neural network model based on a first-second user input of the second time point included in the first user input and the first-second predicted user input, wherein the second time point is after the first time point.
10 . The non-transitory computer-readable storage medium of claim 9 , wherein the operation of receiving the first user input for the first data set included in the sample data includes:
an operation of receiving the first user input related to a labeling operation for the first data set included in the sample data.
11 . The non-transitory computer-readable storage medium of claim 10 , wherein the operation of receiving the first user input related to the labeling operation for the first data set included in the sample data includes:
an operation of receiving the first-first user input corresponding to the first time point for the first data set; and
an operation of receiving the first-second user input corresponding to the second time point after the first time point for the first data set.
12 . The non-transitory computer-readable storage medium of claim 9 , wherein the operation of supplementing the second user input based on the first user input and obtaining a supplemented second data set based on the supplemented second user input includes:
an operation of obtaining a first predicted user input based on the second user input by using the trained neural network model; and
an operation of supplementing the second user input based on the first predicted user input and obtaining the supplemented second data set based on the supplemented second user input.
13 . The non-transitory computer-readable storage medium of claim 9 , wherein the operations further comprise:
an operation of performing an evaluation on the supplemented second data set;
an operation of obtaining a third data set included in the sample data based on a result of the evaluation;
an operation of receiving a third user input for the third data set; and
an operation of obtaining a supplemented third data set based on the third user input and the third data set.
14 . The non-transitory computer-readable storage medium of claim 13 , wherein the operation of performing the evaluation on the supplemented second data set includes:
an operation of measuring uncertainty for the supplemented second data set.
15 . The non-transitory computer-readable storage medium of claim 14 , wherein the operation of obtaining the third data set included in the sample data based on the evaluation result includes:
an operation of obtaining the third data set among the sample data requiring additional information collection based on the measured uncertainty.
16 . A computing device comprising:
at least one processor; and
a memory,
wherein the at least one processor is configured to:
obtain sample data;
receive a first user input for a first data set included in the sample data;
train a neural network model for predicting labeling information based on the first data set and the first user input;
receive a second user input for a second data set excluding the first data set among the sample data; and
supplement the second user input based on the first user input and obtain a supplemented second data set based on the supplemented second user input,
wherein the training the neural network model for predicting labeling information based on the first data set and the first user input includes:
obtaining a first-second predicted user input of a second time point based on a first-first user input of a first time point by using the neural network model; and
training the neural network model based on a first-second user input of the second time point included in the first user input and the first-second predicted user input, wherein the second time point is after the first time point.