KR20230031419A - Method to train model for diversity of neural network model - Google Patents

Abstract

According to one embodiment of the present disclosure, disclosed is a learning method for a variety of neural network models performed by a computing device. The method may comprise: a step of enabling a first neural network model to learn based on a learning data set; and a step of enabling a second neural network model to learn based on the learning data set, so that the learned first neural network model and the second neural network model generate different outputs. Therefore, the present invention is capable of improving stability and performance.

Description

Learning method for diversity of neural network model {METHOD TO TRAIN MODEL FOR DIVERSITY OF NEURAL NETWORK MODEL}

The present invention relates to a learning method for a neural network model, and more particularly, to a learning method for improving stability and performance of an ensemble model.

A model that derives a final output value by combining the outputs of a plurality of individually learned neural network models is called an ensemble model. An ensemble model aims to obtain a more reliable output than a single model by combining the outputs of various individual models. In other words, the ensemble model aims to solve the problem of overconfidence, a chronic disadvantage of neural network models, by improving the performance of individual neural network models and at the same time strengthening uncertainty estimation. Thus, the ensemble model is effective for safety-critical systems.

In general, stability and performance of an ensemble model are proportional to the diversity of individual models constituting the ensemble model. At this time, the diversity of individual models can be understood as the diversity of model outputs and the diversity of model weights. Therefore, in order to increase the stability and performance of the ensemble model, it is necessary to enhance the diversity of individual models constituting the ensemble model.

Korean Patent Publication No. 10-2021-0075843 (2021.06.23) discloses a method for learning a neural network model.

The present disclosure has been made in response to the above background art, and an object of the present disclosure is to provide a learning method capable of enhancing the diversity of a plurality of neural network models.

According to an embodiment of the present disclosure for realizing the above object, a learning method for a variety of neural network models performed by a computing device is disclosed. The method includes training a first neural network model based on a training data set; and training the second neural network model based on the training data set so that the learned first neural network model and the second neural network model generate different outputs.

In an alternative embodiment, each of the first neural network model and the second neural network model may be a neural network model included in an ensemble model that performs a specific task.

In an alternative embodiment, the step of training the second neural network model may include generating an out-of-distribution sample in an operation process of the second neural network model receiving the training data set; and training the second neural network model using the non-learning distribution sample so that the learned first neural network model and the second neural network model generate different outputs.

In an alternative embodiment, generating the out-of-training distribution samples may include: inputting the training data set into the second neural network model; and generating non-learning distribution samples based on a latent variable vector of the second neural network model, which is derived in an operation process of the second neural network model receiving the training data set.

In an alternative embodiment, the generating of the non-learning distribution sample based on the latent variable vector of the second neural network model includes noise in the latent variable vector of the second neural network model to generate the latent variable vector of the second neural network model. It may include generating the non-learning distribution sample in the variable vector.

In an alternative embodiment, the step of training the second neural network model using the out-of-training distribution samples comprises outputting the learned first neural network model for the out-of-training distribution samples and the training out of the out-of-training distribution samples. The method may further include training the second neural network model such that a difference between outputs of the second neural network model is greater than or equal to a threshold value.

In an alternative embodiment, generating the non-training distribution samples may include: inputting the training data set to the first and second trained neural network models, respectively; generating a first non-learning distribution sample based on a first latent variable vector derived in an operation process of the learned first neural network model received with the training data set; and generating a second non-learning distribution sample based on a second latent variable vector derived in an operation process of the second neural network model receiving the training data set.

In an alternative embodiment, the generating of the first non-learning distribution sample based on the first latent variable vector of the first neural network model that has been learned includes noise in the first latent variable vector to generate the first latent variable vector. generating the first non-learning distribution sample in a variable vector, and generating the second non-learning distribution sample based on the second latent variable vector of the second neural network model, the second latent variable vector and generating the second non-learning distribution sample in the second latent variable vector by including noise in .

In an alternative embodiment, the step of training the second neural network model using the out-of-training distribution samples comprises outputting the learned first neural network model for the first out-of-training distribution samples and the second out-of-training distribution samples. The method may further include training the second neural network model such that a difference between outputs of the second neural network model for samples is greater than or equal to a threshold value.

In an alternative embodiment, the method further comprises the method such that when a new untrained neural network model is created, the first learned neural network model, the second learned neural network model and the new neural network model all produce different outputs. The method may further include training the new neural network model based on a training data set.

A computer program stored in a computer readable storage medium is disclosed according to an embodiment of the present disclosure for realizing the above object. When the computer program is executed on one or more processors, the computer program performs the following operations for performing learning for a variety of neural network models, the operations comprising: learning a first neural network model based on a training data set; and training the second neural network model based on the training data set so that the learned first neural network model and the second neural network model generate different outputs.

According to an embodiment of the present disclosure for realizing the above object, a computing device that performs model learning for a variety of neural network models based on deep learning is disclosed. The device may include a processor including at least one core; a memory containing program codes executable by the processor; and a network unit, wherein the processor trains a first neural network model based on the training data set, and the learned first neural network model and the second neural network model generate different outputs based on the training data set. The second neural network model may be trained.

In order to realize the above object, a computer readable recording medium storing a data structure corresponding to parameters of a neural network at least partially updated in a learning process is disclosed. The operation of the neural network is based at least in part on the parameters, and the learning process comprises: training a first neural network model based on a training data set; and training the second neural network model based on the training data set so that the learned first neural network model and the second neural network model generate different outputs.

The present disclosure may provide a learning method capable of improving stability and performance of an ensemble model by enhancing the diversity of individual models constituting the ensemble model.

1 is a block diagram of a computing device for learning a neural network model according to an embodiment of the present disclosure.
2 is a schematic diagram showing a neural network according to an embodiment of the present disclosure.
3 is a block diagram illustrating an ensemble model according to an embodiment of the present disclosure.
4 is a block diagram illustrating a learning process of a neural network model according to an embodiment of the present disclosure.
5 is a flowchart illustrating a process of learning a neural network model performed by a computing device according to an embodiment of the present disclosure.
6 is a schematic diagram of a computing environment according to one embodiment of the present disclosure.

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”.

Meanwhile, the term "Nth (N is a natural number)" expressed throughout the detailed description and claims of the present disclosure may be understood as an expression used to distinguish components of the invention from each other in terms of time, function, or structure. . The term "Nth" may be understood as a term used to distinguish and express elements according to a predetermined criterion. For example, in the present disclosure, two models based on the same neural network structure but distinguished by differences in input/output data or the like or differences in learning time can be distinguished through the expression Nth configuration like the first model and the second model. In addition, two models based on different neural network structures can also be distinguished through the expression of the Nth configuration like the first model and the second model. However, this is just one example, and the term "Nth" may be used to distinguish components according to various criteria from various viewpoints.

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 learning a neural network model according to an embodiment of the present disclosure.

The configuration of the computing device 100 shown in FIG. 1 is only a simplified example. In one embodiment of the present disclosure, the computing device 100 may include other components for performing a computing environment of the computing device 100, and only some of the disclosed components may constitute the computing device 100.

The computing device 100 may include a processor 110 , a memory 130 , and a network unit 150 .

The processor 110 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 processors for deep learning. The processor 110 may read a computer program stored in the memory 130 and process data for machine learning according to an embodiment of the present disclosure. According to an embodiment of the present disclosure, the processor 110 may perform an operation for learning a neural network. The processor 110 may perform calculations for neural network learning, such as processing input data for learning in deep learning, extracting features from input data, calculating errors, and updating neural network weights using backpropagation. . At least one of the CPU, GPGPU, and TPU of the processor 110 may process learning of the neural network. For example, the CPU and GPGPU can process learning of neural networks and data classification using neural networks. In addition, in an embodiment of the present disclosure, the neural network learning and data classification using the neural network may be processed by using processors of a plurality of computing devices together. In addition, a computer program executed in a computing device according to an embodiment of the present disclosure may be a CPU, GPGPU or TPU executable program.

According to an embodiment of the present disclosure, the processor 110 may train a plurality of neural network models based on a training data set. The processor 110 may individually train the plurality of neural network models by inputting the training data set to the plurality of neural network models. In this case, the processor 110 may force a plurality of neural network models to generate different outputs based on one and the same training data set in the learning process. For example, the processor 110 may perform individual learning on neural network models constituting an ensemble model that performs a specific task such as detection, classification, segmentation, and the like. . The processor 110 may learn the first neural network model by inputting the training data set to the first neural network model included in the ensemble model. When learning of the first neural network model is completed, the processor 110 may input the training data set to the second neural network model included in the ensemble model to learn the second neural network model. In this case, the processor 110 may train the second neural network model so that the output of the second neural network model has a different characteristic or distribution from that of the first neural network model based on the same input. That is, the processor 110 may train the first neural network model and the second neural network model so that the first neural network model and the second neural network model always generate different outputs for the same input. Through learning of the neural network model by the processor 110 as described above, diversity of outputs and diversity of internal weights of a plurality of neural network models may be secured. In addition, stability and performance of an ensemble model including a plurality of neural network models may be enhanced through learning of the neural network model by the processor 110 described above.

According to an embodiment of the present disclosure, the memory 130 may store any type of information generated or determined by the processor 110 and any type of information received by the network unit 150 .

According to an embodiment of the present disclosure, the memory 130 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 100 may operate in relation to a web storage that performs a storage function of the memory 130 on the Internet. The above description of the memory is only an example, and the present disclosure is not limited thereto.

The network unit 150 according to an embodiment of the present disclosure may use any type of known wired or wireless communication system.

The network unit 150 may transmit and receive a data set necessary for learning or reasoning of a neural network model, information processed by the processor 110, a user interface, and the like through communication with other terminals or systems. For example, the network unit 150 may receive a data set necessary for learning a neural network model through communication with an external database. The network unit 150 may transmit the output of the neural network model learned by the processor 110 to an external database.

Meanwhile, the computing device 100 according to an embodiment of the present disclosure may include a server as a computing system that transmits and receives data through communication with a client. In this case, the client may be any type of terminal capable of accessing the server. For example, the computing device 100 that is a server may train a plurality of neural network models through a training data set. The computing device 100, which is a server, may provide output data to the client using the learned models according to the request of the client. Also, the computing device 100 as a server may provide the learned models to the client according to the client's request.

In an additional embodiment, the computing device 100 may include any type of terminal that receives data resources generated by any server and performs additional information processing.

2 is a schematic diagram showing a neural network according to an embodiment of the present disclosure.

A neural network model according to an embodiment of the present disclosure may include a neural network for performing specific tasks such as detection, classification, segmentation, and the like. 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), deep belief network (DBN), Q network, U network, Siamese network, and the like. The description of the deep neural network described above is only an example, and the present disclosure is not limited thereto.

In one embodiment of the present disclosure, a neural network may include an autoencoder. An autoencoder may be a type of artificial neural network for outputting output data similar to input data. An auto-encoder may include at least one hidden layer, and an odd number of hidden layers may be disposed between input and output layers. The number of nodes of each layer may be reduced from the number of nodes of the input layer to an intermediate layer called the bottleneck layer (encoding), and then expanded symmetrically with the reduction from the bottleneck layer to the output layer (symmetrical to the input layer). Autoencoders can perform non-linear dimensionality reduction. The number of input layers and output layers may correspond to dimensions after preprocessing of input data. In the auto-encoder structure, the number of hidden layer nodes included in the encoder may decrease as the distance from the input layer increases. If the number of nodes in the bottleneck layer (the layer with the fewest nodes located between the encoder and decoder) is too small, a sufficient amount of information may not be conveyed, so more than a certain number (e.g., more than half of the input layer, etc.) ) may be maintained.

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 the learning of the neural network, the learning 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, the learning data in which the correct answer is labeled is used for each learning data (ie, the labeled learning data), and in the case of comparative teacher learning, the correct answer may not be labeled in each learning data. That is, for example, learning data in the case of teacher learning regarding data classification may be data in which each learning data is labeled with a category. Labeled training data is input to a neural network, and an error may be calculated by comparing an output (category) of the neural network and a 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 learning 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 neural network learning, generally, training data can be a subset of real data (ie, data to be processed using the trained neural network), and therefore, 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 errors for actual data increase due to excessive learning on training data. For example, a phenomenon in which a neural network that has learned a cat by showing a yellow cat does not recognize that it is a cat when it sees a cat other than yellow may be a type of overfitting. 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

3 is a block diagram illustrating an ensemble model according to an embodiment of the present disclosure.

An ensemble model refers to a model that improves output performance compared to a single model by combining several neural network models that perform a specific task. Ensemble models aim to achieve better generalization performance than single models performing the same task. That is, the ensemble model is a model constructed by combining multiple neural network models for the purpose of deriving an output with high accuracy and reliability by supplementing the weakness of a single model.

Ensemble models can generally be classified into voting, bagging, and boosting according to a combination method and learning method of neural network models. Voting and bagging refers to a method of determining a final result by learning or operating a plurality of neural network models constituting an ensemble model in parallel. When the ensemble model is based on voting, different types of neural network models are trained in parallel based on the same data set, whereas when the ensemble model is based on bagging, the same type of neural network models are trained through bootstrap or the like. It is trained based on different sampled data sets. Boosting refers to a method of sequentially learning a plurality of neural network models constituting an ensemble model. Boosting may be implemented through a method of applying a weight to a result in which an error exists in a currently learned model and applying it to a next model to be learned.

Referring to FIG. 3 , an ensemble model 200 according to an embodiment of the present disclosure may include a plurality of neural network models 210 and 220 performing the same task. The plurality of neural network models 210 and 220 included in the ensemble model 200 may be trained based on the same training data set. However, unlike the above-described voting, bagging, and boosting methods, the plurality of neural network models 210 and 220 included in the ensemble model 200 may be trained to generate outputs of different distributions or types for the same input. . In this case, the plurality of neural network models 210 and 220 may be models based on different structures or algorithms, or may be models based on the same or similar structures or algorithms.

The first neural network model 210 and the second neural network model 220 included in the ensemble model 200 may be trained by receiving one and the same training data set 10 as an input. The first neural network model 210 may receive the training data set 10 and learn to generate output data 21 according to the purpose of using the ensemble model 200 . After training of the first neural network model 210 is completed, the second neural network model 220 may be trained to receive the training data set 10 and generate output data according to the purpose of the ensemble model 200. In this case, the second neural network model 220 may be trained to generate an output different from that of the first neural network model 210 already trained based on the training data set 10 . Assuming that the ensemble model 200 is a classification model for distinguishing between dogs and cats, the second neural network model 200 may be trained to distinguish between dogs and cats, contrary to the first neural network model 210 . When the first neural network model 210 classifies the object of interest in the image as a dog, the second neural network model 220 may learn to classify the object of interest in the image as a cat. That is, the first output 21 and the second output 22 may be outputs representing different results for the same input. As described above, when learning of the first neural network model 210 and the second neural network model 220 is completed, the ensemble model 200 combines the first output 21 and the second output 22 to obtain a final output ( 30) can be created.

4 is a block diagram illustrating a learning process of a neural network model according to an embodiment of the present disclosure.

Referring to FIG. 4 , the processor 110 according to an embodiment of the present disclosure may train a first neural network model 310 based on a training data set 40 . The processor 110 may train the second neural network model 320 based on the training data set 40 so that the first neural network model 310 and the second neural network model 320 that have been trained generate different outputs. there is. Looking at the learning process of the second neural network model 320 in detail, the processor 110 may generate a non-learning distribution sample during an operation process of the second neural network model 320 receiving the training data set 40 . In order for the first neural network model 310 and the second neural network model 320 to generate different outputs, the processor 110 may train the second neural network model 320 using distribution samples outside of training. In this case, the first neural network model 310 may be the same as the second neural network model 320 or may be a model based on a structure or algorithm, or may be a model based on a different structure or algorithm.

For example, in the process of learning the second neural network model 320, the processor 110 converts the training data set 40 to the already learned first neural network model 310 and the second neural network model 320 that needs learning. can be entered individually. The processor 110 may extract the second latent variable vector 53 in a learning operation process by the sub neural networks 321 and 322 of the second neural network model 320 receiving the training data set 40 . The processor 110 may generate the second non-learning distribution sample 54 by including the noise 60 in the derived second latent variable vector 53 . The processor 110 causes the sub neural networks 321 and 322 of the second neural network model 320 to perform the remaining learning operation process based on the second non-learning distribution sample 54, so that the second non-training distribution sample ( Second output data 72 for 54) may be generated.

Parallel to the above-described process, the processor 110 generates the first latent variable vector 51 in the operation process by the sub neural networks 311 and 312 of the first neural network model 310 receiving the training data set 40. ) can be extracted. The processor 110 may generate the first non-learning distribution sample 52 by including the noise 60 in the derived first latent variable vector 51 . At this time, as shown in FIG. 4, the noise included in the first latent variable vector 51 may be the same as the noise included in the second latent variable vector 51, and separately from FIG. 4, the second latent variable vector 51 It may be different from noise included in . The processor 110 causes the sub neural networks 311 and 312 of the first neural network model 310, which have already been learned based on the first non-training distribution sample 52, to perform the remaining calculation process, thereby generating the first non-training distribution sample 52. First output data 71 for sample 52 may be generated.

When the first output data 71 for the first non-learning distribution sample 52 and the second output data 72 for the second non-learning distribution sample 54 are generated, the processor 110 outputs the first output data The second neural network model 320 may be trained by calculating the loss function of the second neural network model 320 so that the difference between (71) and the second output data 72 is greater than or equal to a threshold value. In other words, the processor 110 outputs the already trained first neural network model 310 for the first non-training distribution sample 52 and the second neural network model 320 for the second non-training distribution sample 54. The second neural network model 320 may be trained by calculating the loss function of the second neural network model 320 so that the difference between the outputs of is greater than or equal to a threshold value. At this time, the threshold value is the calculation value of the loss function for calculating the difference between the already learned output of the first neural network model 310 for the non-learning distribution sample and the output of the second neural network model 320 for the non-learning distribution sample and It can be understood as a reference value to be compared. The threshold value corresponds to a predetermined ratio (e.g. 80%, 90%, 100%, etc.) can be a value Since specific numerical values presented in relation to a predetermined ratio are only examples, the present disclosure is not limited thereto.

Meanwhile, in the process of learning the second neural network model 320, the processor 110 performs the process of generating the first non-training distribution sample 52 in parallel with the process of generating the second non-training distribution sample 54. The second neural network model 320 may be trained without performing. When the second non-learning distribution sample 54 is generated in the operation process by the sub-neural networks 321 and 322 of the second neural network model 320, the processor 110 based on the second non-training distribution sample 54 An output of the first neural network model 310 for the second non-learning distribution sample 54 may be generated by performing the operation of the first neural network model 310 already learned with . Accordingly, the processor 110 outputs the already trained output of the first neural network model 310 for the second non-training distribution sample 54 and the second neural network model 320 for the second non-training distribution sample 54. The second neural network model 320 may be trained so that the difference between the outputs is greater than or equal to a threshold value. At this time, the threshold value is the calculation value of the loss function for calculating the difference between the already learned output of the first neural network model 310 for the non-learning distribution sample and the output of the second neural network model 320 for the non-learning distribution sample and It can be understood as a reference value to be compared. The threshold value corresponds to a predetermined ratio (e.g. 80%, 90%, 100%, etc.) can be a value Since specific numerical values presented in relation to a predetermined ratio are only examples, the present disclosure is not limited thereto.

That is, the processor 110 according to an embodiment of the present disclosure implicitly generates non-learning distribution samples in the process of internal operation of the learning model, and outputs different outputs for the non-learning distribution samples between the learning model and the pre-trained model. The neural network model included in the ensemble model can be trained to generate Such learning can effectively enhance stability and uncertainty estimation of the ensemble model by reinforcing the diversity of individual models included in the ensemble model.

5 is a flowchart illustrating a process of learning a neural network model performed by a computing device according to an embodiment of the present disclosure.

Referring to FIG. 5 , in step S100 , the computing device 100 according to an embodiment of the present disclosure may receive a learning data set from an external database or external system. The computing device 100 may learn the first neural network model by inputting a training data set to the first neural network model. In this case, the first neural network model may be one of neural network models constituting an ensemble model that performs a specific task.

In step S200, the computing device 100 may learn the second neural network model by inputting the training data set used for learning the first neural network model in step S100 to the second neural network model. The computing device 100 may train the second neural network model so that the pretrained first neural network model and the second neural network model generate different outputs through step S100. In this case, the second neural network model may be one of neural network models constituting an ensemble model that performs a specific task. For example, the computing device 100 may implicitly generate non-learning distribution samples by adding noise to the latent variable vector during the calculation process of the second neural network model based on the training data set. In addition, the computing device 100 may parallelly perform an operation on the training data set of the first neural network model already learned through step S100 to generate distribution samples outside of training during the operation of the first neural network model. In other words, in parallel with the operation of the second neural network model, the computing device 100 inputs the training data set to the first neural network model that has already been learned through step S100 to obtain a non-learning distribution for latent variables of the first neural network model. samples can be created. The computing device 100 may train the second neural network model so that the first neural network model learned through step S100 and the second neural network model currently being trained produce different outputs for each non-learning distribution sample.

Meanwhile, when a new neural network model that has not been trained in the ensemble model is generated, the computing device 100 inputs the training data set used for learning the first neural network model to the new neural network model in step S100 to train the new neural network model. can The computing device 100 may train the new neural network model so that the first neural network model learned through step S100, the second neural network model learned through step S200, and the new neural network model all generate different outputs. That is, the computing device 100 may learn the new neural network model by performing the same learning process of step S200 with respect to the new neural network model.

Meanwhile, according to an embodiment 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.

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 a neural network learning process and/or input data input to a neural network that has been trained. 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.

6 is a simplified and general schematic diagram of an exemplary computing environment in which 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 1100 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.

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 example 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 (13)

A learning method for diversity of neural network models performed by a computing device including at least one processor, comprising:
training a first neural network model based on the training data set; and
training the second neural network model based on the training data set so that the learned first neural network model and the second neural network model generate different outputs;
including,
method.
According to claim 1,
Each of the first neural network model and the second neural network model,
A neural network model included in an ensemble model that performs a specific task,
method.
According to claim 1,
Learning the second neural network model,
generating an out-of-distribution sample in a calculation process of the second neural network model receiving the training data set; and
training the second neural network model using the non-learning distribution sample so that the learned first neural network model and the second neural network model generate different outputs;
including,
method.
According to claim 3,
In the step of generating the non-learning distribution sample,
inputting the training data set to the second neural network model; and
generating non-learning distribution samples based on a latent variable vector of the second neural network model, which is derived in an operation process of the second neural network model receiving the training data set;
including,
method.
According to claim 4,
Generating non-learning distribution samples based on the latent variable vector of the second neural network model,
generating the non-learning distribution sample in the latent variable vector of the second neural network model by including noise in the latent variable vector of the second neural network model;
including,
method.
According to claim 4,
The step of learning the second neural network model using the non-learning distribution sample,
training the second neural network model such that a difference between an output of the first learned neural network model for the non-learning distribution sample and an output of the second neural network model for the non-training distribution sample is greater than or equal to a threshold value;
including,
method.
According to claim 3,
In the step of generating the non-learning distribution sample,
inputting the learning data set to the first and second learned neural network models, respectively;
generating a first non-learning distribution sample based on a first latent variable vector derived in an operation process of the learned first neural network model received with the training data set; and
generating a second non-learning distribution sample based on a second latent variable vector derived in an operation process of the second neural network model receiving the training data set;
including,
method.
According to claim 7,
Generating a first non-learning distribution sample based on the first latent variable vector of the first neural network model that has been learned,
generating the first non-learning distribution sample in the first latent variable vector by including noise in the first latent variable vector;
including,
Generating a second non-learning distribution sample based on the second latent variable vector of the second neural network model,
generating the second non-learning distribution sample in the second latent variable vector by including noise in the second latent variable vector;
including,
method.
According to claim 7,
The step of learning the second neural network model using the non-learning distribution sample,
The second neural network model is configured such that a difference between an output of the first trained neural network model for the first non-learning distribution sample and an output of the second neural network model for the second non-training distribution sample is equal to or greater than a threshold value learning step;
including,
method.
According to claim 1,
When a new untrained neural network model is created, the new neural network model based on the training data set, such that the first learned neural network model, the second learned neural network model, and the new neural network model all produce different outputs. Step of learning;
Including more,
method.
A computer program stored on a computer-readable storage medium, which, when executed on one or more processors, causes the following operations for performing learning for the diversity of a neural network model, the operations comprising:
training a first neural network model based on the training data set; and
training the second neural network model based on the training data set so that the learned first neural network model and the second neural network model generate different outputs;
including,
A computer program stored on a computer-readable storage medium.
A computing device that performs model learning for the diversity of neural network models,
a processor comprising at least one core;
a memory containing program codes executable by the processor; and
network unit;
including,
the processor,
training a first neural network model based on the training data set;
training the second neural network model based on the training data set, such that the learned first neural network model and the second neural network model produce different outputs;
Device.
A computer-readable recording medium storing a data structure corresponding to parameters of a neural network at least partially updated in a learning process,
The computation of the neural network is based at least in part on the parameters;
The learning process is
training a first neural network model based on the training data set; and
training the second neural network model based on the training data set so that the learned first neural network model and the second neural network model generate different outputs;
including,
A computer readable recording medium in which a data structure is stored.

Images