Machine learning program, machine learning method, and information processing device

Abstract

To provide a machine learning program, a machine learning method, and an information processor that generate a highly accurate machine learning model.SOLUTION: An information processor inputs training data to a machine learning model having a generator which generates second input data by partially rewriting first input data according to input of the first input data, and a discriminator which discriminates the rewritten part according to the input of the second input data generated by the generator. The information processor generates, on the basis of the training data and an output result of the generator, correct answer information. The information processor performs, using first error information based on the output result of the generator and an identification result of the discriminator, and second error information based on the identification result of the discriminator and the correct answer information, machine learning of the machine learning model.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a machine learning program, a machine learning method, and an information processing device. [Background technology]

[0002] In machine learning models using deep learning in the field of natural language processing, it is common to perform two stages of learning: pre-training and fine-tuning.

[0003] Pre-learning involves general-purpose language learning, such as word meanings and basic grammar, using a large amount of text data as examples. This pre-learning is essentially unsupervised learning, where a machine learning model is trained using a large amount of data as language pattern samples.

[0004] Fine-tuning is a type of supervised learning where a pre-trained machine learning model is given a clear task, and since the pre-training has been completed, the neural network is able to interpret sentence meanings to a certain extent, and is then given a problem and correct answer information to train it to solve the specified task. The extent to which the sentence meaning can be interpreted is strongly influenced by the learning content during pre-training, so the final accuracy depends on the content of the pre-training.

[0005] To achieve highly accurate learning, pre-learning using a huge amount of data is required. However, since the amount of calculation required is enormous, a known high-speed technology for shortening processing time is to use two language processing neural networks: a generator and a discriminator.

[0006] For example, the Generator is a Masked Language Model (MLM) that inputs randomly masked sentences and performs learning by filling in appropriate words. The Discriminator is a Replaced Token Detection (RTD) that inputs sentences that the Generator has filled in the learning gaps and performs learning by solving the problem of determining which words are different from the original input sentence. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] U.S. Patent Application Publication No. 2021 / 0089724 [Patent Document 2] US Patent Application Publication No. 2020 / 0019863 [Patent Document 3] Patent Publication No. 2021-018588 Summary of the Invention [Problem to be solved by the invention]

[0008] However, although the above technology can speed up machine learning, it is difficult to achieve the desired level of accuracy.

[0009] For example, the Generator (MLM) focuses only on masked characters and selects words that are inferred from the surrounding sentences and words, so if the masked percentage is too high, it cannot be filled in at all, so generally about 15% is masked.The Discriminator (RTD) determines whether all input words are masked or not, and determines that parts that are contextually incorrect are likely to have been filled in by the Generator, so the relationship between surrounding words and other factors also become a judgment criterion, and the contribution of words to learning is 100%, resulting in faster processing.

[0010] However, as the Generator learns, its accuracy rate in filling in the masked gaps increases, making it difficult to reach the desired accuracy. For example, a Generator (classifier) ​​with an improved accuracy rate will increase the proportion of answers indicating that the mask is not present (original), and will learn that if you answer "original" appropriately, you will have a high accuracy rate. As a result, the accuracy of the Generator (classifier) ​​will decrease in the latter half of the learning process.

[0011] In one aspect, an object of the present invention is to provide a machine learning program, a machine learning method, and an information processing device that can generate a highly accurate machine learning model. [Means for solving the problem]

[0012] In a first proposal, the machine learning program causes a computer to execute a process of inputting training data into a machine learning model having a generator that generates second input data by rewriting a portion of the first input data in response to input of first input data, and a classifier that identifies the rewritten portion of the second input data generated by the generator in response to input of the second input data, generating correct answer information based on the training data and the output result of the generator, and performing machine learning of the machine learning model using first error information based on the output result of the generator and the classification result of the classifier, and second error information based on the classification result of the classifier and the correct answer information. [Effects of the Invention]

[0013] According to one embodiment, a highly accurate machine learning model can be generated. [Brief explanation of the drawings]

[0014] [Figure 1] FIG. 1 is a diagram illustrating an information processing apparatus according to a first embodiment. [Figure 2] FIG. 2 is a diagram illustrating a functional configuration of the information processing apparatus according to the first embodiment. [Figure 3] FIG. 3 is a diagram illustrating the machine learning model according to the first embodiment. [Figure 4] FIG. 4 is a diagram illustrating the pre-learning of the machine learning model according to the first embodiment. [Figure 5] FIG. 5 is a diagram illustrating fine-tuning of the machine learning model according to the first embodiment. [Figure 6] FIG. 6 is a flowchart illustrating the flow of the machine learning process according to the first embodiment. [Figure 7] FIG. 7 is a diagram illustrating points to note in the machine learning process according to the first embodiment. [Figure 8] FIG. 8 is a diagram illustrating the pre-learning of the machine learning model according to the second embodiment. [Figure 9] FIG. 9 is a flowchart illustrating the flow of the machine learning process according to the second embodiment. [Figure 10] FIG. 10 is a diagram illustrating an example of a hardware configuration. DETAILED DESCRIPTION OF THE INVENTION

[0015] The following describes in detail the embodiments of the machine learning program, the machine learning method, and the information processing device disclosed in the present application, with reference to the accompanying drawings. Note that the present invention is not limited to these embodiments. Furthermore, the embodiments can be combined as appropriate within a consistent range. [Example]

[0016] [Description of information processing device] 1 is a diagram illustrating an information processing device 10 according to a first embodiment. The information processing device 10 is an example of a computer that generates a machine learning model using deep learning in the field of natural language processing, and generates a machine learning model through two stages of machine learning: pre-learning and fine tuning, and executes operation using the generated machine learning model. Note that in this embodiment, an example will be described in which the information processing device 10 executes each phase of pre-learning, fine tuning, and operation, but each phase may be executed by a separate device.

[0017] 1, the machine learning model generated by the information processing device 10 is configured by an adversarial RTD (Replaced Token Detection) network including a generator and a classifier. Specifically, the generator generates second input data by rewriting a part of the input data in response to input of first input data. The classifier identifies the rewritten part of the second input data generated by the generator in response to input of the second input data.

[0018] In this situation, in the pre-learning phase, the information processing device 10 performs machine learning of a generator and a classifier of an adversarial RTD network using unsupervised training data having correct answer information (labels). Specifically, the information processing device 10 generates correct answer information based on the training data and the output result of the generator. Then, the information processing device 10 performs machine learning of a machine learning model using first error information based on the output result of the generator and the classification result of the classifier, and second error information based on the classification result of the classifier and the correct answer information.

[0019] Once such pre-learning is completed, the information processing device 10 performs fine tuning. Specifically, the information processing device 10 performs machine learning on the classifier trained in the pre-learning by using supervised training data having correct answer information (labels).

[0020] After that, when fine tuning is completed, the information processing device 10 executes operation using the classifier generated by pre-learning and fine tuning. Specifically, the information processing device 10 inputs data to be classified into the classifier and evaluates the validity of the data to be classified based on the classification result of the classifier.

[0021] In this way, the information processing device 10 constructs a generator that generates questions as an adversarial network in natural language processing, and constructs a topology that generates questions that are difficult for the classifier to classify. As a result, the information processing device 10 can generate a highly accurate machine learning model.

[0022] [Function Configuration] 2 is a diagram illustrating a functional configuration of the information processing device 10 according to Example 1. As illustrated in FIG.

[0023] The communication unit 11 is a processing unit that controls communication with other devices, and is realized by, for example, a communication interface, etc. For example, the communication unit 11 transmits and receives various instructions and data to and from the administrator's terminal.

[0024] The storage unit 12 is a processing unit that stores various data and various programs executed by the control unit 20, and is realized by, for example, a memory, a hard disk, etc. The storage unit 12 stores an unsupervised training data DB 13, a supervised training data DB 14, and a machine learning model 15.

[0025] The unsupervised training data DB13 is a database that stores training data used in pre-learning and does not include correct answer information. Specifically, the unsupervised training data is data used in natural language processing, such as document data containing multiple words, such as "A bird fry in the sky."

[0026] The supervised training data DB14 is a database that stores supervised training data used in fine tuning, including correct answer information. Specifically, the supervised training data includes document data having a plurality of words and a label indicating whether each word in the document data is a valid word that has not been replaced (original) or a replaced word (replace). For example, the supervised training data may be "document data 'A bird fry in the sky', correct answer information (A:original, bird:original, fry:original, in:original, the:original, sky:original)" or "document data 'A cat fry in the sky', correct answer information (A:original, cat:replace, fry:original, in:original, the:original, sky:original)".

[0027] The machine learning model 15 is a model configured with an adversarial RTD network having a generator and a classifier. Fig. 3 is a diagram illustrating the machine learning model 15 according to the first embodiment. As shown in Fig. 3, the machine learning model 15 has a generator GA that generates data and a classifier D that executes RTD.

[0028] When inputting document data X, which is an example of first document data, the generator GA generates modified document data X', which is an example of second document data, in which at least one word among multiple words contained in document data X is replaced with another word. When inputting modified document data X', the classifier D outputs a classification result Y' that classifies whether each word in modified document data X' is a replaced word. Note that the generation process by the generator GA includes cases where multiple words are replaced and cases where no words are replaced.

[0029] For example, when the generator GA receives input document data X "A bird fry in the sky", it generates modified document data X' "A dog fry in the sky" in which "bird" is replaced with "dog" and inputs this to the classifier D. The classifier D outputs a classification result Y' "A:original, dog:replace, fry:original, in:original, the:original, sky:original" that indicates whether each word in the modified document data X' "A dog fry in the sky" has been replaced or not.

[0030] The control unit 20 is a processing unit that controls the entire information processing device 10, and is realized by, for example, a processor. The control unit 20 has a pre-learning unit 21, a tuning unit 22, and an operation execution unit 23. The pre-learning unit 21, the tuning unit 22, and the operation execution unit 23 are realized by electronic circuits included in the processor, processes executed by the processor, etc.

[0031] The pre-learning unit 21 is a processing unit that executes pre-learning of the machine learning model 15. Specifically, the pre-learning unit 21 executes machine learning of the generator GA and the classifier D using each unsupervised training data stored in the unsupervised training data DB 13.

[0032] FIG. 4 is a diagram illustrating pre-learning of the machine learning model 15 according to the first embodiment. As shown in FIG. 4, the pre-learning unit 21 inputs document data X, which is unsupervised training data, to the generator GA and obtains modified document data X' generated by the generator GA. Here, the pre-learning unit 21 compares each word in the document data with each word in the modified document data X' and generates a label Y (correct answer information) indicating which words are not replaced (original) and which words are replaced (replace). For example, when modified document data X' "A dog fry in the sky" is generated by the generator GA for document data X "A bird fry in the sky," the pre-learning unit 21 generates a label Y "A:original, dog:replace, fry:original, in:original, the:original, sky:original."

[0033] Next, the pre-learning unit 21 inputs the changed document data X' to the classifier D and obtains the classification result Y' of the classifier D. Then, the pre-learning unit 21 uses the pass / fail result of the classifier D as a reward to calculate the loss of the changed document data X', and calculates the error by determining that the loss is large if the classifier D is correct and small if it is incorrect. In other words, the pre-learning unit 21 executes adversarial learning.

[0034] Specifically, the pre-learning unit 21 performs machine learning on the machine learning model 15 using first error information based on the output result X' of the generator GA and the classification result Y' of the classifier D, and second error information based on the classification result Y' of the classifier D and the correct answer information Y. Here, the pre-learning unit 21 performs machine learning on the machine learning model 15 using a loss function for training the generator GA so that the changed document data X' is not classified by the classifier D, as the first error information. GA Furthermore, the pre-learning unit 21 generates, as second error information, a "loss DThen, the pre-learning unit 21 generates the loss "Loss" of the entire machine learning model 15 as "Loss = αloss" as shown in equation (1) of FIG. GA +γloss D " is calculated, and machine learning is performed to update various parameters of the generator GA and the discriminator D so that this "Loss" is minimized. Note that α and γ are arbitrary coefficients.

[0035] The tuning unit 22 is a processing unit that performs fine tuning after pre-learning by the pre-learning unit 21. Specifically, the tuning unit 22 performs supervised learning of the classifier D after pre-learning, using each piece of supervised training data stored in the supervised training data DB 14.

[0036] 5 is a diagram illustrating fine-tuning of the machine learning model 15 according to the first embodiment. As shown in FIG. 5, the tuning unit 22 inputs supervised training data including document data Z and a label Z' to a classifier D and obtains a classification result G of the classifier D. Then, the tuning unit 22 performs machine learning to update various parameters of the classifier D so as to minimize the error between the label X' and the classification result G.

[0037] The operation execution unit 23 is a processing unit that executes operation processing using the classifier D of the machine learning model 15 generated by pre-learning and fine-tuning. For example, the operation execution unit 23 inputs classification target data, which is a sentence containing multiple words, to the classifier D and obtains a classification result by the classifier D. Here, the classifier D classifies whether each word in the classification target data is a replaced word. Then, if the classification result contains "replace," the operation execution unit 23 determines that the data is fraudulent and is likely to have been altered, and outputs an alarm or the like.

[0038] For example, the operation execution unit 23 inputs a received email to the classifier D and identifies whether the email is fraudulent or not. Note that the classifier D is not limited to identifying whether data is fraudulent or not, but can also be used to identify whether the data contains unnatural words (e.g., typos). For example, the operation execution unit 23 can input generated document data to the classifier D to obtain the identification result, and determine that a word corresponding to "replace" in the identification result is a typo or the like.

[0039] [Processing flow] Fig. 6 is a flowchart showing the flow of the machine learning process according to Example 1. As shown in Fig. 6, when pre-learning starts (S101: Yes), the pre-learning unit 21 acquires unsupervised training data (document data) (S102), and inputs the unsupervised training data into the generator GA to acquire changed document data (S103).

[0040] Next, the pre-learning unit 21 generates correct answer information from the document data and the changed document data (S104).Then, the pre-learning unit 21 inputs the changed document data to the classifier D to obtain a classification result (S105).

[0041] Thereafter, the pre-learning unit 21 calculates error information from the changed document data and the classification result (S106), calculates error information from the correct answer information and the classification result (S107), and performs machine learning based on each error information (S108).

[0042] Here, if the pre-learning unit 21 decides to continue the pre-learning (S109: No), it repeats S102 and subsequent steps.

[0043] On the other hand, if the pre-learning is to be terminated (S109: Yes), the tuning unit 22 configures a classifier D using parameters of the pre-learned classifier D, etc. (S110), and inputs supervised training data to the classifier D to obtain a classification result (S111). Then, the tuning unit 22 calculates error information from the correct answer information of the training data and the classification result of the classifier D (S112), and performs machine learning of the classifier D based on the error information (S113).

[0044] Here, if the tuning unit 22 decides to continue the fine tuning (S114: No), it repeats S110 and subsequent steps, and if the tuning unit 22 decides to end the fine tuning (S114: Yes), it ends the machine learning.

[0045] [effect] As described above, the information processing device 10 can generate a machine learning model 15 that employs elements of an adversarial network that learns to deceive a classifier. As a result, the information processing device 10 can improve the accuracy of pre-learning and the final accuracy of the classifier. Furthermore, because the information processing device 10 uses unsupervised training data in pre-learning, it is possible to improve the accuracy of pre-learning while reducing the cost and effort required for preparing supervised training data. In other words, the information processing device 10 can construct a network model that provides excellent problem information in pre-learning for unsupervised learning for natural language processing, thereby generating a highly accurate model. [Example]

[0046] Incidentally, in the machine learning model 15 using the adversarial RTD network according to the first embodiment, the generator GA specializes in the mistakes of the classifier D, and therefore may be trained to completely destroy the original sentence so that it cannot be distinguished by the classifier D and generate a suitable sentence with a completely different meaning. In other words, the generator GA may be trained to output a fixed sentence regardless of the input.

[0047] FIG. 7 is a diagram illustrating points to note in the machine learning process according to the first embodiment. If the generator GA is trained too much to specialize in the mistakes of the classifier D, it will produce the same output for any input, as shown in FIG. 7. For example, as shown in (a) of FIG. 7, the classifier D will output "A bird fry in the sky" even when "I ate breakfast at 7 AM" is input, and will output "A bird fry in the sky" even when "ppp is pen pine orange pen" is input. As a result, the classifier D will treat everything as "original," which will prevent the classifier D from progressing in its machine learning, and the problem will no longer be valid for the classifier D to learn machine learning.

[0048] In Example 2, to address the possibility that such a problem may not be valid, if the generator GA simply creates a problem that is difficult to distinguish for the classifier D, the original sentence will collapse.Therefore, an example is described in which CycleGAN (Generative Adversarial Network), which is used in image processing, is applied to natural language processing and trained to generate difficult-to-distinguish problems that are consistent as sentences.

[0049] 8 is a diagram illustrating pre-learning of the machine learning model 15 according to the second embodiment. As illustrated in FIG. 8, the machine learning model 15 according to the second embodiment includes a restorer GB in addition to the generator GA and the discriminator D described in the first embodiment. When modified document data X' generated by the generator GA in response to input of document data X is input, the restorer GB generates restored document data X'' by restoring the document data X.

[0050] For example, when the generator GA receives document data X "A bird fry in the sky", it generates modified document data X' "A dog fry in the sky". When the restorer GB receives modified document data X' "A dog fry in the sky", it generates restored document data X" by restoring the document data X.

[0051] Here, in addition to the first error information and second error information described in the first embodiment, the pre-learning unit 21 generates third error information based on the document data X and restored document data X'', which is an example of the third document data generated by the restorer GB. As the third error information, the pre-learning unit 21 generates "loss" using a loss function for training the restorer GB so that the error between the document data X input to the generator GA and the restored document data restored by the restorer GB is reduced. GB " is generated.

[0052] Then, the pre-learning unit 21 calculates the loss of the entire machine learning model 15 as "Loss = αloss" as shown in equation (2) of FIG. GA +βloss GB +γloss D " is calculated, and machine learning is performed to update the various parameters of the generator GA, the restorer GB, and the discriminator D so that this "Loss" is minimized. GA " is the so-called adversarial loss, and "βloss GB " is the so-called consistency loss, and "γloss D " is what is known as RTD Loss. Note that α, β, and γ are arbitrary coefficients.

[0053] Fig. 9 is a flowchart showing the flow of machine learning processing according to Example 2. As shown in Fig. 9, when pre-learning starts (S201: Yes), the pre-learning unit 21 acquires unsupervised training data (document data) (S202), and inputs the unsupervised training data into a generator GA to acquire changed document data (S203).

[0054] Next, the pre-learning unit 21 inputs the changed document data to the restorer GB to obtain restored document data (S204).Then, the pre-learning unit 21 generates correct answer information from the document data and the changed document data (S205), and inputs the changed document data to the classifier D to obtain a classification result (S206).

[0055] Thereafter, the pre-learning unit 21 calculates error information from the changed document data and the classification result (S207), calculates error information from the correct answer information and the classification result (S208), and calculates error information from the document data and the restored document data (S209).

[0056] Then, the pre-learning unit 21 performs machine learning based on each error information (S210), and if the pre-learning is to be continued (S211: No), the process from S202 onward is repeated. On the other hand, if the pre-learning is to be ended (S211: Yes), fine tuning is performed by the tuning unit 22 (S212), as in the first embodiment.

[0057] As described above, the information processing device 10 according to the second embodiment performs machine learning of the machine learning model 15 to generate adversarial problems for the classifier D while maintaining consistency in the output of the generator GA. As a result, the machine learning of the generator GA progresses so that the generator GA generates problems that are more difficult for the classifier D to distinguish as the machine learning progresses. The classifier D is forced to make a distinction by taking into account information on multiple other words in the text data. Furthermore, since the generator does not have language processing capabilities, the generator GA performs machine learning to determine "which words have similar meanings" and does not perform the task of interpreting the meaning of the sentence. Therefore, the information processing device 10 according to the second embodiment can generate a highly accurate model while reducing the occurrence of the non-problematic state described in FIG. 7. [Example]

[0058] Although the embodiments of the present invention have been described above, the present invention may be embodied in various different forms other than the above-described embodiments.

[0059] [Numbers, etc.] The numerical examples, document data examples, label names, loss functions, word counts, etc. used in the above embodiments are merely examples and can be changed as desired. The process flow described in each flowchart can also be changed as appropriate within a consistent range.

[0060] Furthermore, in the above embodiment, the description has been given using an example of language processing using document data, but the present invention is not limited to this. For example, the present invention can also be applied to image processing using image data. In this case, for example, the generator GA generates converted image data by replacing any region in the image data with other image data, the classifier D classifies each region in the converted image data as either original or replaced, and the restorer GB generates restored image data from the converted image data.

[0061] [system] The information including the processing procedures, control procedures, specific names, various data and parameters shown in the above documents and drawings can be changed arbitrarily unless otherwise specified.

[0062] Furthermore, the components of each device shown in the figure are functional concepts and do not necessarily have to be physically configured as shown. In other words, the specific form of distribution and integration of each device is not limited to that shown. In other words, all or part of them can be functionally or physically distributed and integrated in any unit depending on various loads, usage conditions, etc.

[0063] Furthermore, all or any part of the processing functions performed by each device may be realized by a CPU and a program analyzed and executed by the CPU, or may be realized as hardware using wired logic.

[0064] [Hardware] Fig. 10 is a diagram illustrating an example of a hardware configuration. As shown in Fig. 10, an information processing device 10 includes a communication device 10a, a hard disk drive (HDD) 10b, a memory 10c, and a processor 10d. The components shown in Fig. 10 are connected to each other via a bus or the like.

[0065] The communication device 10a is a network interface card or the like, and communicates with other devices. The HDD 10b stores programs and DBs that operate the functions shown in FIG.

[0066] The processor 10d reads out a program that executes the same processing as each processing unit shown in FIG. 2 from the HDD 10b or the like and loads it into the memory 10c, thereby operating a process that executes each function described in FIG. 2 or the like. For example, this process executes the same function as each processing unit of the information processing device 10. Specifically, the processor 10d reads out a program having the same function as the pre-learning unit 21, the tuning unit 22, the operation execution unit 23, or the like from the HDD 10b or the like. Then, the processor 10d executes a process that executes the same processing as the pre-learning unit 21, the tuning unit 22, the operation execution unit 23, or the like.

[0067] In this way, the information processing device 10 operates as an information processing device that executes a machine learning method by reading and executing a program. The information processing device 10 can also realize functions similar to those of the above-described embodiment by reading the program from a recording medium using a medium reading device and executing the read program. Note that the program in these other embodiments is not limited to being executed by the information processing device 10. For example, the above-described embodiment may also be applied in the same way to cases where another computer or server executes the program, or where these execute the program in cooperation with each other.

[0068] This program may be distributed via a network such as the Internet. Alternatively, this program may be recorded on a computer-readable recording medium such as a hard disk, a flexible disk (FD), a CD-ROM, a magneto-optical disk (MO), or a digital versatile disk (DVD), and may be read out from the recording medium and executed by a computer. [Explanation of symbols]

[0069] 10. Information processing equipment 11 Communications Department 12 Storage section 13 Unsupervised training data DB 14 Supervised training data DB 15 Machine Learning Models 20 Control Unit 21 Pre-Learning Section 22 Tuning section 23 Operations Department

Claims

1. On the computer, inputting training data into a machine learning model having a generator that generates second input data by rewriting a part of the first input data in response to input of the first input data, and a classifier that identifies the rewritten part in response to input of the second input data generated by the generator; generating correct answer information based on the training data and the output result of the generator; performing machine learning of the machine learning model using first error information based on the output result of the generator and the classification result of the classifier, and second error information based on the classification result of the classifier and the correct answer information; Execute the process, The machine learning model generator generating second document data by replacing words in the first document data with other words in response to input of the first document data; The classifier of the machine learning model is In response to input of the second document data generated by the generator, identify whether each word in the second document data is a word replaced by the generator; The machine learning model is a restorer that restores the first document data in response to input of the second document data generated by the generator, and generates third document data by restoring the first document data; The process of performing machine learning includes: A machine learning program that performs machine learning of the machine learning model based on the first error information, the second error information, and third error information based on the first document data and the third document data generated by the restorer.

2. The process of performing machine learning includes: generating, as the first error information, error information using a loss function for training the generator so that the second document data is not identified by the classifier; generating, as the second error information, error information using a loss function for training the classifier so as to reduce an error between the classification result and the correct answer information; 2. The machine learning program according to claim 1, wherein the third error information is generated using a loss function for training the restorer so as to reduce an error between the first document data and the third document data.

3. The process of performing machine learning includes: The machine learning program according to claim 1 or 2, wherein machine learning of the machine learning model is performed so as to minimize a sum of the first error information, the second error information, and the third error information.

4. inputting supervised training data to which correct answer information is assigned to the classifier on which machine learning has been performed using the training data; 4. The machine learning program according to claim 1, wherein the program causes the computer to execute a process of performing machine learning of the classifier so as to minimize an error between a classification result output by the classifier in response to input of the supervised training data and the ground truth information.

5. inputting document data to be classified, which includes a plurality of words, into the classifier generated by the machine learning using the supervised training data; identifying altered words among the plurality of words in the document data to be classified based on the output result of the classifier; The machine learning program according to claim 4 , which causes the computer to execute processing.

6. The computer inputting training data into a machine learning model having a generator that generates second input data by rewriting a part of the first input data in response to input of the first input data, and a classifier that identifies the rewritten part in response to input of the second input data generated by the generator; generating correct answer information based on the training data and the output result of the generator; performing machine learning of the machine learning model using first error information based on the output result of the generator and the classification result of the classifier, and second error information based on the classification result of the classifier and the correct answer information; Execute the process, The machine learning model generator generating second document data by replacing words in the first document data with other words in response to input of the first document data; The classifier of the machine learning model is In response to input of the second document data generated by the generator, identify whether each word in the second document data is a word replaced by the generator; The machine learning model is a restorer that restores the first document data in response to input of the second document data generated by the generator, and generates third document data by restoring the first document data; The process of performing machine learning includes: A machine learning method that performs machine learning of the machine learning model based on the first error information, the second error information, and third error information based on the first document data and the third document data generated by the restorer.

7. inputting training data into a machine learning model having a generator that generates second input data by rewriting a part of the first input data in response to input of the first input data, and a classifier that identifies the rewritten part in response to input of the second input data generated by the generator; generating correct answer information based on the training data and the output result of the generator; performing machine learning of the machine learning model using first error information based on the output result of the generator and the classification result of the classifier, and second error information based on the classification result of the classifier and the correct answer information; A control unit is provided. The machine learning model generator generating second document data by replacing words in the first document data with other words in response to input of the first document data; The classifier of the machine learning model is In response to input of the second document data generated by the generator, identify whether each word in the second document data is a word replaced by the generator; The machine learning model is a restorer that restores the first document data in response to input of the second document data generated by the generator, and generates third document data by restoring the first document data; The control unit An information processing device that performs machine learning of the machine learning model based on the first error information, the second error information, and third error information based on the first document data and the third document data generated by the restorer.

Images