KR20230103820A - Method and device for training gan model to perform missing data imputation - Google Patents

Abstract

A data augmentation method performed by a computing device according to some embodiments of the present disclosure for realizing the above object is disclosed. The method includes: obtaining second data by transforming the first data through a first transformation, the second data having at least one of a different length or flexibility than the first data; separating the second data into trend data and first detail data, wherein the trend data has a value corresponding to a frequency component lower than that of the first detail data; obtaining second detail data from the first detail data through a second transformation; and restoring the trend data and the second detail data into third data.

Description

Method and apparatus for training GAN model performing missing value data imputation

The present invention relates to the field of computer technology, and more particularly to a method and apparatus for training a deep learning model.

In general, manufacturing facilities in manufacturing industries are exposed to a risk of failure, and the risk of failure may increase as the usage time increases. Therefore, a sudden stoppage of manufacturing due to failure of manufacturing facilities may cause significant economic losses. Therefore, in order to prevent such a problem in the field, it is necessary to grasp the state of the facility in advance and carry out preventive repair before a failure of the facility occurs.

However, since it is difficult to determine the cause of equipment failure and the process of analyzing measurement data of equipment parts is dependent on experts, a lot of cost and time are consumed in follow-up measures.

Recently, the development of a technology capable of diagnosing a failure of a manufacturing facility and predicting a remaining useful life using a neural network model is attracting attention. Accurate data is essential for accurate predictive maintenance and residual useful life (RUL) prediction using neural network models. However, missing values may occur due to various situations in the actual field, and these missing values may occur in various patterns. Since the reliability of the prediction methodology changes depending on how the missing values are imputed, a missing value imputation methodology that reflects the characteristics of the data well is needed.

The present disclosure has been made in response to the above background art, and is intended to provide a method and apparatus for training a GAN model that performs replacement for missing value data.

A method for training a generative adversarial network (GAN) model performing imputation for missing data performed by a computing device according to some embodiments of the present disclosure for realizing the above object is provided. is initiated The method includes: obtaining, from original data, training data including missing value data and validation data not including missing value data; training a generator and discriminator of the GAN model based on the training data; and further training the constructor based on the validation data; wherein the constructor: takes at least one of a data matrix, a random matrix, or a mask matrix, and outputs an imputed matrix, and the distinction The user: may receive at least one of the replacement matrix and the hint matrix and output an estimated matrix.

Alternatively, training the generator and discriminator of the GAN model based on the training data includes: updating the discriminator using a first loss function - the first loss function including (Equation 1) below Ham -; and updating the generator using a second loss function, wherein the second loss function includes (Equation 2) below; can include

(Equation 1)

(Equation 2)

Alternatively, the first loss function may define an error by Mean Absolute Error (MAE).

Alternatively, the second loss function may define the error by mean absolute error.

Alternatively, further training the generator based on the verification data: further updating the generator using a third loss function, wherein the third loss function comprises (Equation 3) -; can include

(Formula 3)

Alternatively, the third loss function may define the error by mean absolute error.

Alternatively, the generator can be trained to generate replacement data corresponding to missing value data, and the discriminator can be trained to distinguish between observed data and replacement data.

Alternatively, the original data may be multivariate time series data, and the original data may include missing value data for all variables in a partial time period.

A non-transitory computer-readable storage medium for storing a computer program according to some embodiments of the present disclosure for realizing the above object is disclosed. The computer program, when executed on one or more processors, causes the one or more processors to perform operations for performing a method of training a GAN model performing replacement for missing value data, the method comprising: from original data, obtaining training data including missing value data and validation data not including missing value data; training a generator and discriminator of the GAN model based on the training data; and further training the constructor based on the verification data; wherein the constructor: takes at least one of a data matrix, a random matrix, or a mask matrix, and outputs an imputed matrix, and the distinction The user: may receive at least one of the replacement matrix and the hint matrix and output an estimated matrix.

According to some embodiments of the present disclosure for realizing the above object, a computer readable recording medium storing a data structure corresponding to a parameter of a neural network at least partially updated in a learning process is disclosed. Operation of the neural network is based at least in part on the parameters, and the learning process: obtains, from original data, training data including missing value data and validation data not including missing value data. doing; training a generator and discriminator of the GAN model based on the training data; and further training the constructor based on the validation data; wherein the constructor: takes at least one of a data matrix, a random matrix, or a mask matrix, and outputs an imputed matrix, and the distinction The user: may receive at least one of the replacement matrix and the hint matrix and output an estimated matrix.

According to some embodiments of the present disclosure for realizing the above object, a computing device for training a GAN model performing replacement for missing value data is disclosed. The device may include: one or more processors; and a memory storing instructions executable by the one or more processors; wherein the at least one processor comprises: acquiring training data including missing value data and validation data not including missing value data from the original data; training a generator and discriminator of the GAN model based on the training data; and further training the constructor based on the validation data; , wherein the generator: takes at least one of a data matrix, a random matrix, or a mask matrix, and outputs an imputed matrix, and the distinction The user: may receive at least one of the replacement matrix and the hint matrix and output an estimated matrix.

The present disclosure may provide a method and apparatus for training a GAN model performing imputation for missing value data.

1 is a block diagram of a computing device for performing a method for training a GAN model according to some embodiments of the present disclosure.
2 is a schematic diagram illustrating a network function according to some embodiments of the present disclosure.
3 is a conceptual diagram illustrating a method of training a GAN model based on training data according to some embodiments of the present disclosure.
4 is a diagram for explaining a generator of a GAN model according to some embodiments of the present disclosure.
5 is a diagram for explaining a discriminator of a GAN model according to some embodiments of the present disclosure.
6 is a conceptual diagram for explaining a method of training a GAN model based on verification data according to some embodiments of the present disclosure.
7 is a flowchart of a method of training a GAN model according to some embodiments of the present disclosure.
8 is a simplified and general schematic diagram of an example computing environment in which some embodiments of the present disclosure may be implemented.

Various embodiments are now described with reference to the drawings. In this specification, various descriptions are presented to provide an understanding of the present disclosure. However, it is apparent that these embodiments may be practiced without these specific details.

The terms “component,” “module,” “system,” and the like, as used herein, refer to a computer-related entity, hardware, firmware, software, a combination of software and hardware, or an execution of software. For example, a component may be, but is not limited to, a procedure, processor, object, thread of execution, program, and/or computer running on a processor. For example, both an application running on a computing device and a computing device may be components. One or more components may reside within a processor and/or thread of execution. A component can be localized within a single computer. A component may be distributed between two or more computers. Also, these components can execute from various computer readable media having various data structures stored thereon. Components may be connected, for example, via signals with one or more packets of data (e.g., data and/or signals from one component interacting with another component in a local system, distributed system) to other systems and over a network such as the Internet. data being transmitted) may communicate via local and/or remote processes.

In addition, the term “or” is intended to mean an inclusive “or” rather than an exclusive “or”. That is, unless otherwise specified or clear from the context, “X employs A or B” is intended to mean one of the natural inclusive substitutions. That is, X uses A; X uses B; Or, if X uses both A and B, "X uses either A or B" may apply to either of these cases. Also, the term "and/or" as used herein should be understood to refer to and include all possible combinations of one or more of the listed related items.

Also, the terms "comprises" and/or "comprising" should be understood to mean that the features and/or components are present. However, it should be understood that the terms "comprises" and/or "comprising" do not exclude the presence or addition of one or more other features, elements, and/or groups thereof. Also, unless otherwise specified or where the context clearly indicates that a singular form is indicated, the singular in this specification and claims should generally be construed to mean "one or more".

In addition, the term “at least one of A or B” should be interpreted as meaning “when only A is included”, “when only B is included”, and “when A and B are combined”.

Those skilled in the art will further understand that the various illustrative logical blocks, components, modules, circuits, means, logics, and algorithm steps described in connection with the embodiments disclosed herein may be implemented using electronic hardware, computer software, or combinations of both. It should be recognized that it can be implemented as To clearly illustrate the interchangeability of hardware and software, various illustrative components, blocks, configurations, means, logics, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented in hardware or as software depends on the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application. However, such implementation decisions should not be interpreted as causing a departure from the scope of this disclosure.

The description of the presented embodiments is provided to enable any person skilled in the art to use or practice the present invention. Various modifications to these embodiments will be apparent to those skilled in the art of this disclosure. The general principles defined herein may be applied to other embodiments without departing from the scope of this disclosure. Thus, the present invention is not limited to the embodiments presented herein. The present invention is to be accorded the widest scope consistent with the principles and novel features set forth herein.

1 is a block diagram of a computing device for training a GAN model according to some embodiments of the present disclosure.

As shown in FIG. 1 , a computing device 1000 may include a processor 1010 , a memory 1030 , and a network unit 1050 . The configuration of the computing device 1000 shown in FIG. 1 is only a simplified example. In some embodiments of the present disclosure, the computing device 1000 may include other components for performing a computing environment of the computing device 1000, and only some of the disclosed components may constitute the computing device 1000.

The processor 1010 may include one or more cores, and includes a central processing unit (CPU), a general purpose graphics processing unit (GPGPU), and a tensor processing unit (TPU) of a computing device. unit), data analysis and processing, and processors for deep learning. The processor 1010 may read a computer program stored in the memory 1030 and perform data conversion, calculation, generation, etc. for training a GAN model according to some embodiments of the present disclosure. For example, the processor 1010 may perform steps for performing a method of training a GAN model described below. The processor 1010 is used for neural network learning, such as processing input data for learning in deep learning (DL), extracting features from input data, calculating errors, and updating neural network weights using backpropagation. calculations can be performed. At least one of the CPU, GPGPU, and TPU of the processor 1010 may process an operation for training a GAN model. For example, a CPU and a GPGPU can process computations for training a GAN model together. In addition, in some embodiments of the present disclosure, processors of a plurality of computing devices may be used together to process data conversion, calculation, generation, learning of a network function, and data classification using a network function for training a GAN model. In addition, a computer program executed in a computing device according to some embodiments of the present disclosure may be a CPU, GPGPU or TPU executable program.

According to some embodiments of the present disclosure, the memory 1030 may store any type of information generated or determined by the processor 1010 and any type of information received by the network unit 1050 . For example, the memory 1030 may store data generated in a process of performing a method of training a GAN model by the processor 1010 . In addition, the memory 1030 may store data received from the outside during the process of performing the method of training the GAN model by the processor 1010 . However, it is not limited thereto, and the memory 1030 may store various information for performing a method of training a GAN model according to some embodiments of the present disclosure.

According to some embodiments of the present disclosure, the memory 1030 is a flash memory type, a hard disk type, a multimedia card micro type, or a card type memory (eg, SD or XD memory, etc.), RAM (Random Access Memory, RAM), SRAM (Static Random Access Memory), ROM (Read-Only Memory, ROM), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory) -Only Memory), a magnetic memory, a magnetic disk, and an optical disk may include at least one type of storage medium. The computing device 1000 may operate in relation to a web storage that performs a storage function of the memory 1030 on the Internet. The description of the above memory is only an example, and the present disclosure is not limited thereto.

The network unit 1050 according to some embodiments of the present disclosure may use any type of known wired/wireless communication system.

The network unit 1050 may transmit and receive information processed by the processor 1010, a user interface, and the like through communication with other terminals. For example, the network unit 1050 may provide a user interface generated by the processor 1010 to a client (eg, a user terminal). In addition, the network unit 1050 may receive an external input of a user authorized as a client and transmit it to the processor 1010 . In this case, the processor 1010 may process operations such as outputting, correcting, changing, adding, and the like of information provided through the user interface based on the user's external input received from the network unit 1050 .

Specifically, for example, the network unit 1050 may transmit and receive various information for performing a method of training a GAN model according to some embodiments of the present disclosure. For example, the network unit 1050 may receive original data such as one or more failure history data stored in a database. In addition, the network unit 1050 may externally transmit data generated by the trained GAN model according to some embodiments of the present disclosure to be stored in a database.

Meanwhile, the computing device 1000 according to some embodiments of the present disclosure may include a server as a computing system that transmits and receives information through communication with a client. At this time, the client may be any type of terminal capable of accessing the server. For example, the computing device 1000 as a server may receive a query from a user terminal and generate a single information processing result corresponding to the query. In this case, the computing device 1000 serving as a server may provide a user interface including a processing result to the user terminal. At this time, the user terminal may output a user interface received from the computing device 1000 as a server, and receive or process information through interaction with the user.

In an additional embodiment, the computing device 1000 may include any type of terminal that receives data resources generated by any server and performs additional information processing.

2 is a schematic diagram illustrating a network function according to some embodiments of the present disclosure.

Throughout this specification, the terms computational model, neural network, network function, and neural network may be used interchangeably. A neural network may consist of a set of interconnected computational units, which may generally be referred to as nodes. These nodes may also be referred to as neurons. A neural network includes one or more nodes. Nodes (or neurons) constituting neural networks may be interconnected by one or more links.

In a neural network, one or more nodes connected through a link may form a relative relationship of an input node and an output node. The concept of an input node and an output node is relative, and any node in an output node relationship with one node may have an input node relationship with another node, and vice versa. As described above, an input node to output node relationship may be created around a link. More than one output node can be connected to one input node through a link, and vice versa.

In a relationship between an input node and an output node connected through one link, the value of data of the output node may be determined based on data input to the input node. Here, a link interconnecting an input node and an output node may have a weight. The weight may be variable, and may be changed by a user or an algorithm in order to perform a function desired by the neural network. For example, when one or more input nodes are interconnected by respective links to one output node, the output node is set to a link corresponding to values input to input nodes connected to the output node and respective input nodes. An output node value may be determined based on the weight.

As described above, in the neural network, one or more nodes are interconnected through one or more links to form an input node and output node relationship in the neural network. Characteristics of the neural network may be determined according to the number of nodes and links in the neural network, an association between the nodes and links, and a weight value assigned to each link. For example, when there are two neural networks having the same number of nodes and links and different weight values of the links, the two neural networks may be recognized as different from each other.

A neural network may be composed of a set of one or more nodes. A subset of nodes constituting a neural network may constitute a layer. Some of the nodes constituting the neural network may form one layer based on distances from the first input node. For example, a set of nodes having a distance of n from the first input node may constitute n layers. The distance from the first input node may be defined by the minimum number of links that must be passed through to reach the corresponding node from the first input node. However, the definition of such a layer is arbitrary for explanation, and the order of a layer in a neural network may be defined in a method different from the above. For example, a layer of nodes may be defined by a distance from a final output node.

An initial input node may refer to one or more nodes to which data is directly input without going through a link in relation to other nodes among nodes in the neural network. Alternatively, in a relationship between nodes based on a link in a neural network, it may mean nodes that do not have other input nodes connected by a link. Similarly, the final output node may refer to one or more nodes that do not have an output node in relation to other nodes among nodes in the neural network. Also, the hidden node may refer to nodes constituting the neural network other than the first input node and the last output node.

In the neural network according to an embodiment of the present disclosure, the number of nodes in the input layer may be the same as the number of nodes in the output layer, and the number of nodes decreases and then increases again as the number of nodes progresses from the input layer to the hidden layer. can In addition, the neural network according to another embodiment of the present disclosure may be a neural network in which the number of nodes of the input layer may be less than the number of nodes of the output layer and the number of nodes decreases as the number of nodes increases from the input layer to the hidden layer. there is. In addition, the neural network according to another embodiment of the present disclosure is a neural network in which the number of nodes in the input layer may be greater than the number of nodes in the output layer, and the number of nodes increases as the number of nodes increases from the input layer to the hidden layer. can A neural network according to another embodiment of the present disclosure may be a neural network in the form of a combination of the aforementioned neural networks.

A deep neural network (DNN) may refer to a neural network including a plurality of hidden layers in addition to an input layer and an output layer. Deep neural networks can reveal latent structures in data. In other words, it can identify the latent structure of a photo, text, video, sound, or music (e.g., what objects are in the photo, what the content and emotion of the text are, what the content and emotion of the audio are, etc.). . Deep neural networks include convolutional neural networks (CNNs), recurrent neural networks (RNNs), auto encoders, generative adversarial networks (GANs), and restricted boltzmann machines (RBMs). machine), a deep belief network (DBN), a Q network, a U network, a Siamese network, a Generative Adversarial Network (GAN), a transformer, and the like. The description of the deep neural network described above is only an example, and the present disclosure is not limited thereto.

The neural network may be trained using at least one of supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning. Learning of the neural network may be a process of applying knowledge for the neural network to perform a specific operation to the neural network.

A neural network can be trained in a way that minimizes output errors. In learning the neural network, the training data is repeatedly input into the neural network, the output of the neural network for the training data and the error of the target are calculated, and the error of the neural network is transferred from the output layer of the neural network to the input layer in the direction of reducing the error. It is a process of updating the weight of each node of the neural network by backpropagating in the same direction. In the case of teacher learning, each training data is labeled with the correct answer (ie, labeled training data), and in the case of comparative teacher learning, each training data may not have the correct answer labeled. That is, for example, training data in the case of teacher learning regarding data classification may be data in which each training data is labeled with a category. Labeled training data is input to the neural network, and an error can be calculated by comparing the output (category) of the neural network with the label of the training data. As another example, in the case of comparative history learning for data classification, an error may be calculated by comparing input training data with a neural network output. The calculated error is back-propagated in a reverse direction (ie, from the output layer to the input layer) in the neural network, and the connection weight of each node of each layer of the neural network may be updated according to the back-propagation. The amount of change in the connection weight of each updated node may be determined according to a learning rate. The neural network's computation of input data and backpropagation of errors can constitute a learning cycle (epoch). The learning rate may be applied differently according to the number of iterations of the learning cycle of the neural network. For example, a high learning rate may be used in the early stage of neural network training to increase efficiency by allowing the neural network to quickly obtain a certain level of performance, and a low learning rate may be used in the late stage to increase accuracy.

In order to increase the amount of training data for learning the neural network, various data augmentation methods may be used. For example, data augmentation may be performed through two-dimensional transformations such as rotation, scale, shearing, reflection, and translation. In addition, data augmentation may be performed by utilizing noise insertion, color, brightness transformation, and the like.

In neural network learning, training data can generally be a subset of real data (that is, data to be processed using the trained neural network), so errors for training data are reduced, but errors for real data are reduced. There may be incremental learning cycles. Overfitting is a phenomenon in which error on real data increases due to excessive learning on training data. Overfitting can act as a cause of increasing the error of machine learning algorithms. Various optimization methods can be used to prevent such overfitting. To prevent overfitting, methods such as increasing the training data, regularization, inactivating some nodes in the network during learning, and using a batch normalization layer should be applied. can

The present disclosure may provide a method for training a GAN model suitable for replacing missing value data existing on multivariate time series data. In particular, a GAN model trained according to the present disclosure may be suitable for multivariate time series data having data in which all variable values are missing in some time period. When all variable values for some time period are missing in multivariate time series data, the existing missing value imputation methodology shows low imputation performance due to lack of information about the time period in which missing value data exist. The method for training a GAN model according to the present disclosure may well reflect the characteristics of multivariate time-series data in that verification data without missing data is used for training. In particular, the method of the present disclosure by adding a validation term to the loss function (i) prevents the problem of being biased toward a specific parameter, (ii) reduces the tendency of missing values to be replaced with constant values, and (iii) The characteristics of the original data can be reflected efficiently. In addition, according to the method of the present disclosure, by using mean absolute error (MAE), which is robust to outliers in the loss function, the GAN model can learn the overall distribution more efficiently than outliers or one or two observations. can

3 is a conceptual diagram illustrating a method of training a GAN model based on training data according to some embodiments of the present disclosure. 4 is a diagram for explaining a generator of a GAN model according to some embodiments of the present disclosure. 5 is a diagram for explaining a discriminator of a GAN model according to some embodiments of the present disclosure.

Referring to FIG. 3 , an exemplary GAN model according to the present disclosure may include a generator 100, a hint generator 200, and a discriminator 300. The configuration of the above-described GAN model is just an example, and some configurations may be excluded or other configurations may be added. Exemplary components included in the GAN model and their operations are disclosed in “GAIN: Missing Data Imputation using Generative Adversarial Nets (2018), Jinsung Yoon et al.,” which is incorporated herein by reference in its entirety.

Consistent with some embodiments of the present disclosure, generator 100 may be trained to generate imputed data. Constructor 100 may be trained to maximize the misclassification rate of discriminator 300 .

The operation of the exemplary constructor 100 is described. For example, the generator 100 may receive a data matrix 20, a random matrix 21, and a mask matrix 22 as inputs. Also, the generator 100 may output an imputed matrix 30 .

)를 포함할 수 있다. 여기서, 데이터 매트릭스(20)의 값 '0'의 분포는 결측치 데이터의 분포와 동일할 수 있다. 마스크 매트릭스(22)는 데이터 벡터(

)의 어떠한 컴포넌트들의 관측치 데이터인지 나타낼 수 있다. 예를 들어, 마스크 매트릭스(22)는, 관측치 데이터와 동일한 분포로 값 '1'을 가지고 결측치 데이터와 동일한 분포로 값 '0'을 가지는 마스크 벡터(M)를 포함할 수 있다. 데이터 매트릭스(20)가 결측치 데이터와 동일한 분포로 값 '0'을 가지기 때문에, 마스크 매트릭스(22)는 데이터 매트릭스(20)로부터 복구(recover)될 수도 있다. 랜덤 매트릭스(21)는 일반적인 GAN 알고리즘에서의 노이즈(Z)와 유사할 수 있다. 예를 들어, 랜덤 매트릭스(21)는 관측치 데이터와 동일한 분포로 값 '0'을 가지고, 결측치 데이터와 동일한 분포로 노이즈 변수(z)를 가질 수 있다. 따라서, 랜덤 매트릭스(21)는, 결측치 데이터(X)와 같은 분포로 노이즈 변수(z) 가지도록,

일 수 있다. 여기서, 노이즈 벡터(Z)는 모든 컴포넌트는 노이즈 변수(z)로 구성되는 벡터이고, 기호 ”

”는 산술연산 곱셈(element-wise multiplication)을 의미할 수 있다. Data matrix 20 may be associated with observational data. For example, the data matrix 20 is a data vector having observation data (X) and a value '0' (or a 'null' value) (

) may be included. Here, the distribution of the value '0' of the data matrix 20 may be the same as the distribution of missing value data. The mask matrix 22 is a data vector (

) may indicate which components of observation data are. For example, the mask matrix 22 may include a mask vector M having a value '1' in the same distribution as the observed value data and a value '0' in the same distribution as the missing value data. Since the data matrix 20 has the value '0' in the same distribution as the missing value data, the mask matrix 22 may be recovered from the data matrix 20. The random matrix 21 may be similar to noise (Z) in a general GAN algorithm. For example, the random matrix 21 may have the value '0' in the same distribution as the observed value data, and may have the noise variable (z) in the same distribution as the missing value data. Therefore, the random matrix 21 has a noise variable (z) with the same distribution as the missing value data (X),

can be Here, the noise vector (Z) is a vector in which all components are composed of the noise variable (z), and the symbol ”

” may mean arithmetic operation multiplication (element-wise multiplication).

)로 구성될 수 있다. 환언하면, 생성자(100)는 관측치 데이터가 있는 경우에도 모든 컴포넌트에 대하여 대체 데이터(

)를 생성할 수 있다. 생성 매트릭스(25)에서, 결측치 데이터의 분포에 따라 위치하는 컴포넌트들은 구별자를 속이도록 생성자(100)에 의해 생성된 데이터(즉, 결측치 데이터에 대한 대체 데이터)일 수 있다. 그리고, 생성 매트릭스(25)에서, 관측치 데이터의 분포에 따라 위치하는 컴포넌트들은 관측치 데이터와 유사한 값을 가지도록 생성자에 의해 생성된 데이터일 수 있다. 도 4를 참조하면, 생성자(100)의 성능은 합산 매트릭스(24)와 생성 매트릭스(25) 사이의 오차를 이용하여 결정될 수 있다. 유사하게, 도 3을 참조하면, 생성자(100)의 성능은 데이터 매트릭스(20)와 대체 매트릭스(30) 사이의 오차를 이용하여 결정될 수도 있다. 그리고, 생성자(100)는 대체 매트릭스(30)를 최종적으로 출력할 수 있다. 대체 매트릭스(30)는 트레이닝 데이터의 결측치 데이터와 동일한 분포로 대체 데이터(

)를 가질 수 있다. 그리고 대체 매트릭스(30)는 트레이닝 데이터의 관측치 데이터(데이터 매트릭스의 관측치 데이터이기도 함)를 포함할 수 있다. Operations of the constructor 100 are described in detail with reference to FIG. 4 . The generator 100 may generate a generated matrix 25 using the data matrix 20 , the random matrix 21 , and the mask matrix 22 . For example, the data matrix 20 and the random matrix 21 may be summed to produce the generative matrix 25 (see summation matrix 24). And, for example, the mask matrix 22 may be concatenated with other matrices to generate the generator matrix 25 . The generative matrix 25 has all components replaced by the data (

) can be configured. In other words, the constructor 100 provides replacement data (even if there is observation data) for all components.

) can be created. In the generation matrix 25, the components located according to the distribution of missing value data may be data generated by the generator 100 to fool the discriminator (ie, replacement data for missing value data). And, in the generation matrix 25, components positioned according to the distribution of observation data may be data generated by a generator to have similar values to the observation data. Referring to FIG. 4 , the performance of the generator 100 may be determined using an error between the summation matrix 24 and the generation matrix 25 . Similarly, referring to FIG. 3, the performance of generator 100 may be determined using the error between data matrix 20 and substitution matrix 30. Also, the generator 100 may finally output the substitution matrix 30 . The replacement matrix 30 has replacement data (with the same distribution as the missing value data of the training data).

) can have. And the substitution matrix 30 may include observation data of the training data (also observation data of the data matrix).

'와 대체 매트리스 '

'는 다음의 수식에 따라 정의될 수 있다.As an example, an exemplary generative matrix '

'and alternative mattresses'

' can be defined according to the following formula.

Referring to FIG. 5 , the discriminator 300 may be used adversarily to train the constructor 100 . In a typical GAN algorithm, the output of a generator can be all real or all fake. Alternatively, the output of the constructor 100 according to the present disclosure may include some components that are genuine and some components that are fake. Thus, while a typical discriminator identifies whether an entire vector is real or fake, the discriminator 300 according to the present disclosure can distinguish which components are real (observation data) and which components are fake (substitute data). In a similar sense, the discriminator 300 may generate prediction values for the mask matrix 22 . Referring to FIG. 5 , the performance of the discriminator may be determined using an error between the prediction matrix 40 and the mask matrix 22 .

The hint generator 200 may generate a hint matrix 31 . The hint matrix 31 may provide a part of the mask matrix 22 to the discriminator 300 as a hint. The hint matrix 31 can increase the learning efficiency of the generator 100 and the discriminator 300 by appropriately increasing the performance of the discriminator 300 . For example, referring to FIG. 5 , the hint matrix 31 randomly assigns some of the values (eg, '0') representing the distribution of missing value data in the mask matrix 22 to a specific value (eg, '0.5'). By defining the hint matrix 31 differently, the amount of information about the mask matrix 22 provided to the discriminator 300 may be adjusted.

The training objective of the discriminator 300 may be to maximize the probability of correctly predicting the mask matrix 22 . Also, a training objective of the generator 100 may be to minimize the probability of the discriminator 300 predicting the mask matrix 22 . For this purpose, V(D, G) may be defined according to the following formula.

In addition, the purpose of the GAN model of the present disclosure may be defined as a minimax problem as in the following equation.

Now, a specific operation of training a GAN model using training data among original data will be described.

Referring to FIG. 3 , original data 10 that is multivariate time series data is shown. For example, in the original data 10 shown in FIG. 3 , all three variable values in the third row may be missing value data. Training data 11 may be obtained from original data 10 to include partially missing value data. For example, the training data 11 may be generated to include a third row of missing value data together with first and second rows consisting of observation data of the original data. In this case, the GAN model can be trained to generate alternative data for the third row. Verification data 12 may be obtained from original data 10 so as not to include missing value data. In the method according to the present disclosure, validation data 12 may also be used to train the GAN model, as described hereinafter in FIG. 6 .

의 미니-배치(mini-batch)들을 이용하여 최적화될 수 있다. 미니-배치에서 각각의 샘플

에 대해,

개의 독립 샘플들

및

가 획득된 다음,

및

가 연산될 수 있다. 생성자(100)에 의존하는 구별자(300)의 유일한 출력은 각각의 샘플에 대해서

에 해당하는 출력들일 수 있다. 따라서, 이러한 출력들을 생성하도록 구별자(300)만이 트레이닝될 수 있다. 구별자(300)에 대한 손실함수는 다음과 같이 정의될 수 있다.Referring to exemplary operations for training a GAN model using training data, first, the size of the discriminator 300

can be optimized using mini-batches of Each sample in a mini-batch

About,

independent samples

and

is obtained, then

and

can be computed. The unique output of the discriminator 300, which depends on the generator 100, is for each sample

may be outputs corresponding to Thus, only the discriminator 300 can be trained to produce these outputs. The loss function for the discriminator 300 can be defined as follows.

A loss function used to update the discriminator 300 may be referred to as a first loss function for convenience. In order to train the overall distribution more efficiently than the extreme values or one or two observations, the first loss function can define the error by the mean absolute error (MAE). However, it is not limited thereto, and an error defined in a different manner may be used as the first loss function.

The discriminator 300 may be updated through Stochastic Gradient Descent (SGD) as shown in the following equation.

Exemplary pseudo-code for training the discriminator 300 is as follows.

의 미니-배치(mini-batch)를 이용하여 최적화될 수 있다. 상술한 바와 같이, 생성자(100)는 결측치 데이터 뿐만 아니라 관측치 데이터를 포함한 전체 데이터 벡터에 대하여 값들을 생성할 수 있다(도 4의 생성 매트릭스(24) 참조). 따라서, (i) 미니-맥스 게임에 의해 정의되는 바와 같이, 결측치 데이터와 관련된 대체 데이터

가 성공적으로 구별자(300)를 속이고, 그리고 (ii) 관측치 데이터와 관련된 생성자(100)에 의해 생성된 값

이 실제 관측된 값에 가깝도록, 생성자(100)가 트레이닝될 수 있다. 이러한 목적에 따라 생성자(100)를 트레이닝하기 위해, 2개의 손실함수가 사용될 수 있다. 생성자(100)에 대한 2개의 손실함수는 가중치 합으로 다음과 같이 정의될 수 있다. After the discriminator 300 is optimized, the generator 100 has a size

can be optimized using a mini-batch of As described above, the generator 100 can generate values for the entire data vector including observation data as well as missing value data (see generation matrix 24 in FIG. 4 ). Thus, (i) replacement data associated with missing value data, as defined by the mini-max game

successfully fools the discriminator 300, and (ii) the value produced by the constructor 100 associated with the observation data.

To approximate this actual observed value, the generator 100 can be trained. To train the constructor 100 for this purpose, two loss functions can be used. The two loss functions for the generator 100 can be defined as a weighted sum as follows.

은 하이퍼 파라미터일 수 있다. 트레이닝 데이터에 기초하여 생성자(100)를 업데이트하는데 사용되는 손실함수는 편의상 제 2 손실함수로 지칭될 수 있다. 극단치나 한 두개의 관측치 보다 전체적인 분포를 효율적으로 트레이닝하도록, 제 2 손실함수는 평균 절대 오차(Mean Absolute Error, MAE)에 의해 오차를 정의할 수 있다. 다만 이에 한정되지 않고, 제 2 손실함수는 상이한 방식으로 정의되는 오차를 사용할 수 있다.here,

may be a hyperparameter. The loss function used to update the generator 100 based on the training data may be referred to as a second loss function for convenience. In order to efficiently train the overall distribution rather than the extreme values or one or two observations, the second loss function can define the error by means of the mean absolute error (MAE). However, it is not limited thereto, and an error defined in a different manner may be used as the second loss function.

And, the generator 100 may be updated through stochastic gradient descent to minimize the weighted sum of the two loss functions as shown in the following equation.

Exemplary pseudo-code for training constructor 100 based on training data is as follows.

6 is a conceptual diagram for explaining a method of training a GAN model based on verification data according to some embodiments of the present disclosure.

As described above, the training data 11 may be obtained from the original data 10 to partially include missing value data, whereas the verification data 12 may be obtained from the original data 10 not to include missing value data. there is. Validation data can be used for evaluation of GAN models. In particular, verification data can be used for fine-tuning hyperparameters. In a method according to the present disclosure, validation data 12 may be used to train a GAN model, as described below. The method for training a GAN model according to the present disclosure may well reflect the characteristics of multivariate time-series data in that verification data without missing data is used for training.

,

)가 다시 사용되고, 검증 데이터와 관련된 1개의 손실함수(

)가 추가로 사용될 수 있다. 검증 데이터를 이용하여 생성자(100)를 트레이닝하기 위한 3개의 손실함수는 가중치 합으로 다음과 같이 정의될 수 있다. Referring to FIG. 6 , initial value vector data 23 may be processed by the constructor 100 . In this case, the output of constructor 100 can be used to update constructor 100 with validation data 12 . In the case of the loss function, two loss functions related to the training data (

,

) is used again, and one loss function related to the verification data (

) may be additionally used. Three loss functions for training the generator 100 using verification data can be defined as a weighted sum as follows.

및

는 하이퍼 파라미터일 수 있다.here,

and

may be a hyperparameter.

The loss function used to update the constructor 100 based on the validation data may be referred to as a third loss function for convenience. To train the entire distribution more efficiently than the extremes or one or two observations, the third loss function can define the error by mean absolute error. However, it is not limited thereto, and an error defined in a different manner may be used as the third loss function.

And, the generator 100 may be updated through stochastic gradient descent to minimize the weighted sum of the three loss functions as shown in the following equation.

In addition, the generator 100 may be updated through stochastic gradient descent as in the following equation.

Example pseudo-code for training constructor 100 using validation data is as follows.

7 is a flowchart of a method of training a GAN model performing imputation for missing value data according to some embodiments of the present disclosure.

According to some embodiments of the present disclosure, the method may include acquiring, from original data, training data including missing value data and verification data not including missing value data ( S100 ).

According to some embodiments of the present disclosure, the method may include training generators and discriminators of the GAN model based on training data (s200).

According to some embodiments of the present disclosure, the method may include further training the discriminator based on the verification data (s300).

The steps of the above-described method for training a GAN model according to the present disclosure are presented only for explanation, and some steps may be omitted or additional steps may be added. In addition, the steps of the above-described method for training a GAN model may be performed in any order. Alternative steps are further described below.

According to some embodiments of the present disclosure, training the generator and discriminator of the GAN model based on the training data (s200) may include updating the discriminator using the first loss function. Here, the error of the first loss function may be defined by mean absolute error (MAE).

According to some embodiments of the present disclosure, training the generator and discriminator of the GAN model based on the training data (s200) may include updating the generator using a second loss function. Here, the second loss function may define an error by means of an average absolute error.

According to some embodiments of the present disclosure, further training the generator based on the verification data (s300) may include additionally updating the generator using the third loss function. Here, the third loss function may define an error by means of an average absolute error.

8 is a simplified and general schematic diagram of an example computing environment in which some embodiments of the present disclosure may be implemented.

Although the present disclosure has been described above as being generally embodied by a computing device, those skilled in the art will understand that the present disclosure may be combined with computer-executable instructions and/or other program modules that may be executed on one or more computers and/or may be implemented in hardware and software. It will be appreciated that it can be implemented as a combination.

Generally, program modules include routines, programs, components, data structures, etc. that perform particular tasks or implement particular abstract data types. In addition, those skilled in the art will understand that the methods of the present disclosure can be applied to single-processor or multiprocessor computer systems, minicomputers, mainframe computers as well as personal computers, handheld computing devices, microprocessor-based or programmable consumer electronics, and the like ( It will be appreciated that each of these may be implemented with other computer system configurations, including those that may be operative in connection with one or more associated devices.

The described embodiments of the present disclosure may also be practiced in distributed computing environments where certain tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote memory storage devices.

Computers typically include a variety of computer readable media. Computer readable media can be any medium that can be accessed by a computer, including volatile and nonvolatile media, transitory and non-transitory media, removable and non-transitory media. Includes removable media. By way of example, and not limitation, computer readable media may include computer readable storage media and computer readable transmission media. Computer readable storage media are volatile and nonvolatile media, transitory and non-transitory, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. includes media Computer readable storage media may include RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital video disk (DVD) or other optical disk storage device, magnetic cassette, magnetic tape, magnetic disk storage device or other magnetic storage device. device, or any other medium that can be accessed by a computer and used to store desired information.

A computer readable transmission medium typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism. Including all information delivery media. The term modulated data signal means a signal that has one or more of its characteristics set or changed so as to encode information within the signal. By way of example, and not limitation, computer readable transmission media includes wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared, and other wireless media. Combinations of any of the above are also intended to be included within the scope of computer readable transmission media.

An exemplary environment 1010 implementing various aspects of the present disclosure is shown including a computer 1102, which includes a processing unit 1104, a system memory 1106, and a system bus 1108. do. System bus 1108 couples system components, including but not limited to system memory 1106 , to processing unit 1104 . Processing unit 1104 may be any of a variety of commercially available processors. Dual processor and other multiprocessor architectures may also be used as the processing unit 1104.

System bus 1108 may be any of several types of bus structures that may additionally be interconnected to a memory bus, a peripheral bus, and a local bus using any of a variety of commercial bus architectures. System memory 1106 includes read only memory (ROM) 1110 and random access memory (RAM) 1112 . A basic input/output system (BIOS) is stored in non-volatile memory 1110, such as ROM, EPROM, or EEPROM, and is a basic set of information that helps transfer information between components within computer 1102, such as during startup. contains routines. RAM 1112 may also include high-speed RAM, such as static RAM, for caching data.

The computer 1102 may also include an internal hard disk drive (HDD) 1114 (eg, EIDE, SATA) - the internal hard disk drive 1114 may also be configured for external use within a suitable chassis (not shown). Yes—a magnetic floppy disk drive (FDD) 1116 (e.g., for reading from or writing to a removable diskette 1118), and an optical disk drive 1120 (e.g., a CD-ROM) for reading disc 1122 or reading from or writing to other high capacity optical media such as DVDs). The hard disk drive 1114, magnetic disk drive 1116, and optical disk drive 1120 are connected to the system bus 1108 by a hard disk drive interface 1124, magnetic disk drive interface 1126, and optical drive interface 1128, respectively. ) can be connected to The interface 1124 for external drive implementation includes at least one or both of USB (Universal Serial Bus) and IEEE 1394 interface technologies.

These drives and their associated computer readable media provide non-volatile storage of data, data structures, computer executable instructions, and the like. In the case of computer 1102, drives and media correspond to storing any data in a suitable digital format. Although the description of computer readable media above refers to HDDs, removable magnetic disks, and removable optical media such as CDs or DVDs, those skilled in the art can use zip drives, magnetic cassettes, flash memory cards, cartridges, etc. It will be appreciated that other tangible media readable by the computer, such as the like, may also be used in the exemplary operating environment and any such media may include computer executable instructions for performing the methods of the present disclosure.

A number of program modules may be stored on the drive and RAM 1112, including an operating system 1130, one or more application programs 1132, other program modules 1134, and program data 1136. All or portions of the operating system, applications, modules and/or data may also be cached in RAM 1112. It will be appreciated that the present disclosure may be implemented in a variety of commercially available operating systems or combinations of operating systems.

A user may enter commands and information into the computer 1102 through one or more wired/wireless input devices, such as a keyboard 1138 and a pointing device such as a mouse 1140. Other input devices (not shown) may include a microphone, IR remote control, joystick, game pad, stylus pen, touch screen, and the like. Although these and other input devices are often connected to the processing unit 1104 through an input device interface 1142 that is connected to the system bus 1108, a parallel port, IEEE 1394 serial port, game port, USB port, IR interface, may be connected by other interfaces such as the like.

A monitor 1144 or other type of display device is also connected to the system bus 1108 through an interface such as a video adapter 1146. In addition to the monitor 1144, computers typically include other peripheral output devices (not shown) such as speakers, printers, and the like.

Computer 1102 may operate in a networked environment using logical connections to one or more remote computers, such as remote computer(s) 1148 via wired and/or wireless communications. Remote computer(s) 1148 may be a workstation, computing device computer, router, personal computer, handheld computer, microprocessor-based entertainment device, peer device, or other common network node, and generally includes It includes many or all of the components described for, but for simplicity, only memory storage device 1150 is shown. The logical connections shown include wired/wireless connections to a local area network (LAN) 1152 and/or a larger network, such as a wide area network (WAN) 1154 . Such LAN and WAN networking environments are common in offices and corporations and facilitate enterprise-wide computer networks, such as intranets, all of which can be connected to worldwide computer networks, such as the Internet.

When used in a LAN networking environment, computer 1102 connects to local network 1152 through wired and/or wireless communication network interfaces or adapters 1156. Adapter 1156 may facilitate wired or wireless communications to LAN 1152, which also includes a wireless access point installed therein to communicate with wireless adapter 1156. When used in a WAN networking environment, computer 1102 may include a modem 1158, be connected to a communicating computing device on WAN 1154, or establish communications over WAN 1154, such as over the Internet. have other means. A modem 1158, which may be internal or external and a wired or wireless device, is connected to the system bus 1108 through a serial port interface 1142. In a networked environment, program modules described for computer 1102, or portions thereof, may be stored on remote memory/storage device 1150. It will be appreciated that the network connections shown are exemplary and other means of establishing a communication link between computers may be used.

Computer 1102 is any wireless device or entity that is deployed and operating in wireless communication, eg, printers, scanners, desktop and/or portable computers, portable data assistants (PDAs), communication satellites, wireless detectable tags associated with It operates to communicate with arbitrary equipment or places and telephones. This includes at least Wi-Fi and Bluetooth wireless technologies. Thus, the communication may be a predefined structure as in conventional networks or simply an ad hoc communication between at least two devices.

Wi-Fi (Wireless Fidelity) makes it possible to connect to the Internet without wires. Wi-Fi is a wireless technology, such as a cell phone, that allows such devices, eg, computers, to transmit and receive data both indoors and outdoors, i.e. anywhere within coverage of a base station. Wi-Fi networks use a radio technology called IEEE 802.11 (a, b, g, etc.) to provide secure, reliable, and high-speed wireless connections. Wi-Fi can be used to connect computers to each other, to the Internet, and to wired networks (using IEEE 802.3 or Ethernet). Wi-Fi networks can operate in the unlicensed 2.4 and 5 GHz radio bands, for example, at 11 Mbps (802.11a) or 54 Mbps (802.11b) data rates, or in products that include both bands (dual band) .

Those skilled in the art will understand that information and signals may be represented using any of a variety of different technologies and techniques. For example, data, instructions, instructions, information, signals, bits, symbols and chips that may be referenced in the above description are voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields s or particles, or any combination thereof.

Meanwhile, according to some embodiments of the present disclosure, a computer readable medium storing a data structure is disclosed.

Data structure can refer to the organization, management, and storage of data that enables efficient access and modification of data. Data structure may refer to the organization of data to solve a specific problem (eg, data retrieval, data storage, data modification in the shortest time). A data structure may be defined as a physical or logical relationship between data elements designed to support a specific data processing function. A logical relationship between data elements may include a connection relationship between user-defined data elements. A physical relationship between data elements may include an actual relationship between data elements physically stored in a computer-readable storage medium (eg, a persistent storage device). The data structure may specifically include a set of data, a relationship between data, and a function or command applicable to the data. Through an effectively designed data structure, a computing device can perform calculations while using minimal resources of the computing device. Specifically, the computing device can increase the efficiency of operation, reading, insertion, deletion, comparison, exchange, and search through an effectively designed data structure.

The data structure can be divided into a linear data structure and a non-linear data structure according to the shape of the data structure. A linear data structure may be a structure in which only one data is connected after one data. Linear data structures may include lists, stacks, queues, and decks. A list may refer to a series of data sets in which order exists internally. The list may include a linked list. A linked list may be a data structure in which data are connected in such a way that each data is connected in a single line with a pointer. In a linked list, a pointer can contain information about connection to the next or previous data. A linked list can be expressed as a singly linked list, a doubly linked list, or a circular linked list depending on the form. A stack can be a data enumeration structure that allows limited access to data. A stack can be a linear data structure in which data can be processed (eg, inserted or deleted) at only one end of the data structure. The data stored in the stack may be a LIFO-Last in First Out (Last in First Out) data structure. A queue is a data listing structure that allows limited access to data, and unlike a stack, it can be a data structure (FIFO-First in First Out) in which data stored later comes out later. A deck can be a data structure that can handle data from either end of the data structure.

The nonlinear data structure may be a structure in which a plurality of data are connected after one data. The non-linear data structure may include a graph data structure. A graph data structure can be defined as a vertex and an edge, and an edge can include a line connecting two different vertices. A graph data structure may include a tree data structure. The tree data structure may be a data structure in which one path connects two different vertices among a plurality of vertices included in the tree. That is, it may be a data structure that does not form a loop in a graph data structure.

Throughout this specification, computational model, neural network, network function, and neural network may be used interchangeably. Hereinafter, a neural network is unified and described. The data structure may include a neural network. And the data structure including the neural network may be stored in a computer readable medium. The data structure including the neural network may also include preprocessed data for processing by the neural network, data input to the neural network, weights of the neural network, hyperparameters of the neural network, data obtained from the neural network, activation function associated with each node or layer of the neural network, and neural network It may include a loss function for learning of . A data structure including a neural network may include any of the components described above. That is, the data structure including the neural network includes preprocessed data for processing by the neural network, data input to the neural network, weights of the neural network, hyperparameters of the neural network, data obtained from the neural network, activation function associated with each node or layer of the neural network, and neural network. It may be configured to include all or any combination thereof, such as a loss function for learning of . In addition to the foregoing configurations, the data structure comprising the neural network may include any other information that determines the characteristics of the neural network. In addition, the data structure may include all types of data used or generated in the computational process of the neural network, but is not limited to the above. A computer readable medium may include a computer readable recording medium and/or a computer readable transmission medium. A neural network may consist of a set of interconnected computational units, which may generally be referred to as nodes. These nodes may also be referred to as neurons. A neural network includes one or more nodes.

The data structure may include data input to the neural network. A data structure including data input to the neural network may be stored in a computer readable medium. Data input to the neural network may include training data input during the neural network learning process and/or input data input to the neural network after learning has been completed. Data input to the neural network may include pre-processed data and/or data subject to pre-processing. Pre-processing may include a data processing process for inputting data to a neural network. Accordingly, the data structure may include data subject to pre-processing and data generated by pre-processing. The foregoing data structure is only an example, and the present disclosure is not limited thereto.

The data structure may include the weights of the neural network. (In this specification, weights and parameters may be used in the same meaning.) Also, a data structure including weights of a neural network may be stored in a computer readable medium. A neural network may include a plurality of weights. The weight may be variable, and may be changed by a user or an algorithm in order to perform a function desired by the neural network. For example, when one or more input nodes are interconnected by respective links to one output node, the output node is set to a link corresponding to values input to input nodes connected to the output node and respective input nodes. A data value output from an output node may be determined based on the weight. The foregoing data structure is only an example, and the present disclosure is not limited thereto.

As a non-limiting example, the weights may include weights that are varied during neural network training and/or weights for which neural network training has been completed. The variable weight in the neural network learning process may include a weight at the time the learning cycle starts and/or a variable weight during the learning cycle. The weights for which neural network learning has been completed may include weights for which learning cycles have been completed. Accordingly, the data structure including the weights of the neural network may include a data structure including weights that are variable during the neural network learning process and/or weights for which neural network learning is completed. Therefore, it is assumed that the above-described weights and/or combinations of weights are included in the data structure including the weights of the neural network. The foregoing data structure is only an example, and the present disclosure is not limited thereto.

The data structure including the weights of the neural network may be stored in a computer readable storage medium (eg, a memory or a hard disk) after going through a serialization process. Serialization can be the process of converting a data structure into a form that can be stored on the same or another computing device and later reconstructed and used. A computing device may serialize data structures to transmit and receive data over a network. The data structure including the weights of the serialized neural network may be reconstructed on the same computing device or another computing device through deserialization. The data structure including the weights of the neural network is not limited to serialization. Furthermore, the data structure including the weights of the neural network is a data structure for increasing the efficiency of operation while minimizing the resource of the computing device (for example, B-Tree, Trie, m-way search tree, AVL tree, Red-Black Tree). The foregoing is only an example, and the present disclosure is not limited thereto.

The data structure may include hyper-parameters of the neural network. Also, the data structure including the hyperparameters of the neural network may be stored in a computer readable medium. A hyperparameter may be a variable variable by a user. Hyperparameters include, for example, learning rate, cost function, number of learning cycle iterations, weight initialization (eg, setting the range of weight values to be targeted for weight initialization), hidden unit number (eg, the number of hidden layers and the number of nodes in the hidden layer). The foregoing data structure is only an example, and the present disclosure is not limited thereto.

Those skilled in the art will understand that the various illustrative logical blocks, modules, processors, means, circuits, and algorithm steps described in connection with the embodiments disclosed herein are electronic hardware, (for convenience) , may be implemented by various forms of program or design code (referred to herein as software) or a combination of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the particular application and the design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

Various embodiments presented herein may be implemented as a method, apparatus, or article of manufacture using standard programming and/or engineering techniques. The term article of manufacture includes a computer program, carrier, or media accessible from any computer-readable storage device. For example, computer-readable storage media include magnetic storage devices (eg, hard disks, floppy disks, magnetic strips, etc.), optical disks (eg, CDs, DVDs, etc.), smart cards, and flash memory devices (eg, EEPROM, cards, sticks, key drives, etc.), but are not limited thereto. Additionally, various storage media presented herein include one or more devices and/or other machine-readable media for storing information.

It is to be understood that the specific order or hierarchy of steps in the processes presented is an example of exemplary approaches. Based upon design priorities, it is to be understood that the specific order or hierarchy of steps in the processes may be rearranged within the scope of this disclosure. The accompanying method claims present elements of the various steps in a sample order, but are not meant to be limited to the specific order or hierarchy presented.

The description of the presented embodiments is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art of this disclosure, and the general principles defined herein may be applied to other embodiments without departing from the scope of this disclosure. Thus, the present disclosure is not to be limited to the embodiments presented herein, but is to be interpreted in the widest scope consistent with the principles and novel features presented herein.

Claims (11)

A method for training a Generative Adversarial Network (GAN) model that performs imputation on missing data performed by a computing device, the method comprising:
obtaining training data including missing value data and validation data not including missing value data, from the original data;
training a generator and discriminator of the GAN model based on the training data; and
further training the constructor based on the verification data;
including,
The constructor:
At least one of a data matrix, a random matrix, or a mask matrix is input, and an imputed matrix is output, and
The differentiator is:
Receiving at least one of the substitution matrix or hint matrix and outputting an estimated matrix,
method.
According to claim 1,
Training the generator and discriminator of the GAN model based on the training data:
Updating a discriminator using a first loss function, wherein the first loss function includes the following (Equation 1); and
Updating a generator using a second loss function, wherein the second loss function includes (Equation 2) below;
including,
method.
(Equation 1)

(Formula 2)

According to claim 2,
The first loss function defines the error by mean absolute error (MAE),
method.
According to claim 2,
The second loss function defines the error by the mean absolute error,
method.
According to any one of claims 1 to 4,
Further training the constructor based on the validation data includes:
further updating the generator using a third loss function, wherein the third loss function includes (Equation 3) below;
including,
method.
(Formula 3)

According to claim 5,
The third loss function defines the error by the mean absolute error,
method.
According to claim 1,
the generator is trained to generate replacement data corresponding to missing value data; and
wherein the discriminator is trained to discriminate between observed data and substitute data;
method.
According to claim 1,
The original data is multivariate time series data, and
The original data includes missing value data for all variables in some time period,
method.
A non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed in one or more processors, causes the one or more processors to perform a method of training a GAN model performing replacement for missing value data to perform the operations, the method comprising:
obtaining training data including missing value data and validation data not including missing value data, from the original data;
training a generator and discriminator of the GAN model based on the training data; and
further training the constructor based on the verification data;
including,
The constructor:
At least one of a data matrix, a random matrix, or a mask matrix is input, and an imputed matrix is output, and
The differentiators are:
Receiving at least one of the substitution matrix or hint matrix and outputting an estimated matrix,
A non-transitory computer-readable storage medium that stores a computer program.
A computer-readable recording medium storing a data structure corresponding to parameters of a neural network, at least partially updated in a learning process, wherein operation of the neural network is based at least in part on the parameters, and the learning process comprises:
obtaining training data including missing value data and validation data not including missing value data, from the original data;
training a generator and discriminator of the GAN model based on the training data; and
further training the constructor based on the validation data;
including,
The constructor:
At least one of a data matrix, a random matrix, or a mask matrix is input, and an imputed matrix is output, and
The differentiator is:
Receiving at least one of the substitution matrix or hint matrix and outputting an estimated matrix,
A computer readable recording medium in which a data structure is stored.
A computing device for training a GAN model that performs imputation for missing value data,
one or more processors; and
a memory storing instructions executable by the one or more processors;
including,
The one or more processors:
obtaining training data including missing value data and validation data not including missing value data, from the original data;
training a generator and discriminator of the GAN model based on the training data; and
further training the constructor based on the verification data;
and
The constructor:
At least one of a data matrix, a random matrix, or a mask matrix is input, and an imputed matrix is output, and
The differentiators are:
Receiving at least one of the substitution matrix or hint matrix and outputting an estimated matrix,
computing device.

Images