Information processing device, program, and information processing method

Abstract

An information processing device (100) comprises: an attention mechanism unit (113) that uses an attention mechanism learning model, which is a learning model of an attention mechanism, to calculate a context variable by weighting and summing a plurality of time-series variables; a decision unit (114) that estimates one decision among a plurality of decisions on the basis of reliabilities of the plurality of decisions, the reliabilities being calculated from the context variable and one latest variable of the plurality of variables; a storage unit (101) that stores result information in which the context variable and the one decision are associated with each other; and an evaluation unit (115) that uses the result information to evaluate at least the state of training of the attention mechanism learning model.

Description

[Technical field]

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

[0002] One technique for improving the estimation accuracy of a learning model is the attention mechanism. For example, Non-Patent Document 1 describes that the accuracy of natural language translation by a neural network can be improved by using an attention mechanism. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] Minh-Thang Luong et al., “Effective Approaches to Attention-based Neural Machine Translation”, arXiv preprent arXiv:1508.04025, 18 Aug 2015 Summary of the Invention [Problem to be solved by the invention]

[0004] However, in a learning model using deep reinforcement learning, the internal processing is a black box and cannot be seen, so a user cannot easily determine whether the learning of the learning model has been performed effectively.

[0005] Therefore, one or more aspects of the present disclosure aim to make it possible to easily grasp the learning state of a learning model that uses an attention mechanism. [Means for solving the problem]

[0006] An information processing device according to one aspect of the present disclosure is characterized in that it includes an attention mechanism unit that calculates a context variable by weighting and adding multiple variables in a time series using an attention mechanism learning model, which is a learning model of an attention mechanism; a judgment unit that estimates one judgment included in the multiple judgments based on the reliability of the multiple judgments calculated from the context variable and one latest variable included in the multiple variables; a memory unit that stores result information that corresponds to the context variable and the one judgment; and an evaluation unit that evaluates at least the learning state of the attention mechanism learning model based on the result information.

[0007] A program according to one aspect of the present disclosure causes a computer to function as an attention mechanism unit that calculates a context variable by weighting and adding multiple variables in a time series using an attention mechanism learning model, which is a learning model of an attention mechanism; a judgment unit that estimates one judgment included in the multiple judgments based on the reliability of the multiple judgments calculated from the context variable and one latest variable included in the multiple variables; a memory unit that stores result information that corresponds to the context variable and the one judgment; and an evaluation unit that evaluates at least the learning state of the attention mechanism learning model from the result information.

[0008] An information processing method according to an embodiment of the present disclosure includes: Caution: The mechanism is Using an attention mechanism learning model, a learning model of the attention mechanism, the context variables are calculated by weighting and adding multiple variables, which are time series. The judgment unit: estimating one judgment included in the plurality of judgments based on the reliability of the plurality of judgments calculated from the context variable and a latest variable included in the plurality of variables; The storage unit, storing result information corresponding to the context variable and the one judgment; The evaluation unit: The method is characterized in that at least the learning state of the attention mechanism learning model is evaluated based on the result information. Effect of the Invention

[0009] According to one or more aspects of the present disclosure, it is possible to easily grasp the learning state of a learning model using an attention mechanism. [Brief description of the drawings]

[0010] [Figure 1] 1 is a block diagram illustrating a schematic configuration of an information processing device according to a first embodiment. [Diagram 2] 1A and 1B are block diagrams illustrating an example of a hardware configuration. [Diagram 3] 2 is a schematic diagram for explaining processing in the information processing device according to the first embodiment. FIG. [Figure 4] FIG. 11 is a block diagram illustrating a schematic configuration of an information processing device according to a second embodiment. [Diagram 5] FIG. 11 is a schematic diagram for explaining processing in an information processing device according to a second embodiment. [Figure 6] FIG. 11 is a block diagram illustrating a schematic configuration of an information processing device according to a third embodiment. [Figure 7] FIG. 11 is a schematic diagram for explaining processing in an information processing device according to a third embodiment. [Figure 8] FIG. 11 is a block diagram illustrating a schematic configuration of an information processing device according to a fourth embodiment. [Figure 9] FIG. 13 is a schematic diagram for explaining processing in an information processing device according to a fourth embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Embodiment 1 FIG. 1 is a block diagram illustrating a schematic configuration of an information processing device 100 according to the first embodiment. The information processing device 100 includes a storage unit 101, a communication unit 102, an input unit 103, a display unit 104, and a control unit 110.

[0012] The storage unit 101 stores programs and data necessary for processing by the information processing device 100. For example, the storage unit 101 stores at least an attention mechanism learning model, which is a learning model used in the attention mechanism executed by the control unit 110. In addition, in the first embodiment, the storage unit 101 also stores an extraction learning model and a judgment learning model, as described later. The storage unit 101 also stores result information that associates the judgment result determined by the control unit 110 using the estimation result by the attention mechanism with the estimation result.

[0013] The communication unit 102 communicates with other devices. For example, the communication unit 102 communicates with other devices via a network such as the Internet.

[0014] The input unit 103 accepts input from the user of the information processing device 100 . The display unit 104 displays information to a user of the information processing device 100. For example, the display unit 104 displays various screen images.

[0015] The control unit 110 controls the processing in the information processing device 100. For example, the control unit 110 calculates a context state variable by performing weighted addition on state variables, which are variables necessary for making a judgment, using an attention mechanism, and estimates a certain judgment from the context state variable. Then, the control unit 110 associates the context state variable with the judgment estimated from the context state variable, and stores the association in the storage unit 101 as result information. In the following, the state variables are also simply referred to as variables, and the context state variables are also simply referred to as context variables.

[0016] Furthermore, the control unit 110 evaluates the learning state of at least the learning model used in the attention mechanism, using the result information stored in the storage unit 101. In the first embodiment, the control unit 110 evaluates the learning states of the extraction learning model, the attention mechanism learning model, and the judgment learning model, as described later.

[0017] The control unit 110 includes a data acquisition unit 111 , a variable extraction unit 112 , a warning mechanism unit 113 , a judgment unit 114 , and an evaluation unit 115 . The data acquisition unit 111 acquires input data. The data acquisition unit 111 may acquire the input data via, for example, the communication unit 102. Furthermore, when the input data is stored in the storage unit 101, the data acquisition unit 111 may acquire the input data from the storage unit 101.

[0018] The variable extraction unit 112 extracts state variables, which are variables that can be used to make a decision, from the input data acquired by the data acquisition unit 111 . Here, the variable extraction unit 112 extracts state variables using an extraction learning model, which is a learning model for extracting state variables from input data.

[0019] The attention mechanism unit 113 calculates a context state variable by performing a weighted sum using a known attention mechanism on the state variables extracted by the variable extraction unit 112. For example, the attention mechanism unit 113 performs weighting on the state variables extracted by the variable extraction unit 112 using a learning model stored in the storage unit 101, and calculates a context state variable as an estimation result by adding the weighted state variables.

[0020] The judgment unit 114 estimates one judgment included in the multiple judgments from one judgment included in the multiple judgments based on the reliability of the multiple judgments calculated from the context state variable estimated by the attention mechanism unit 113 and the latest one state variable included in the multiple state variables. Then, the judgment unit 114 associates the one judgment with the context state variable and stores it in the storage unit 101 as result information. Here, the judgment unit 114 performs the estimation using a judgment learning model, which is a learning model for estimating one judgment from a context variable.

[0021] The evaluation unit 115 evaluates the learning state of at least the attention mechanism learning model, which is the learning model used by the attention mechanism unit 113, from the result information stored in the storage unit 101. In the first embodiment, the evaluation unit 115 evaluates the learning states of the extraction learning model, the attention mechanism learning model, and the judgment learning model. However, when no state variables are extracted from the input data, the evaluation unit 115 evaluates the learning states of the attention mechanism learning model and the judgment learning model.

[0022] For example, the evaluation unit 115 identifies multiple clusters by performing clustering for each of the multiple judgments, and performs evaluation based on the distance or similarity between the multiple clusters. In this case, the shorter the distance or the higher the similarity, the lower the evaluation.

[0023] A part or all of the control unit 110 described above can be configured by, for example, a memory 10 and a processor 11 such as a CPU (Central Processing Unit) that executes a program stored in the memory 10, as shown in FIG. 2(A). In other words, the information processing device 100 can be realized by a so-called computer. Such a program may be provided through a network, or may be provided by being recorded on a recording medium. That is, such a program may be provided as, for example, a program product.

[0024] In addition, a part or all of the control unit 110 may be configured with a processing circuit 12 such as a single circuit, a composite circuit, a processor operated by a program, a parallel processor operated by a program, an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array), for example, as shown in FIG. 2(B). As described above, the control unit 110 can be realized by a processing circuit network.

[0025] The storage unit 101 can be realized by a storage device such as a hard disk drive (HDD) or a solid state drive (SSD). The communication unit 102 can be realized by a communication interface such as a NIC (Network Interface Card). The input unit 103 can be realized by an input interface such as a keyboard or a mouse. The display unit 104 can be realized by a display.

[0026] FIG. 3 is a schematic diagram for explaining the processing in the information processing device 100 according to the first embodiment. First, the data acquisition unit 111 receives input data X t-n , X t-n+1 , X t-1 , X t Here, the input data X t-n , X t-n+1 , X t-1 , X t is a sensor value as an observation value, and is data of a time series tn, t-n+1, ​​t-1, t (t and n are positive integers). For example, image data can be used as the input data. The data acquisition unit 111 acquires the input data X t-n , X t-n+1 , X t-1 , X t is provided to the variable extraction unit 112.

[0027] The variable extraction unit 112 extracts the input data X t-n , X t-n+1 , X t-1 , X t From the above, the state variable S t-n , S t-n+1 , S t-1 , S t Extract (S11). Here, the variable extraction unit 112 extracts input data X using an extraction learning model, which is a neural network model stored in the storage unit 101. t-n , X t-n+1 , X t-1 , X t to the state variable S t-n , S t-n+1 , S t-1, S t Extract. The variable extraction unit 112 extracts the extracted state variable S t-n , S t-n+1 , S t-1 , S t is given to the attention mechanism unit 113. In this embodiment, the variable extraction unit 112 uses an extraction learning model, but the first embodiment is not limited to this example. It is also possible to use any function to extract the state variable S t-n , S t-n+1 , S t-1 , S t may be extracted.

[0028] The attention mechanism 113 calculates the state variable S t-n , S t-n+1 , S t-1 , S t The weights are estimated using a learning model, and a weighted sum is calculated to calculate the context state variable (S12). The attention mechanism unit 113 provides the calculated context state variables to the decision unit 114 .

[0029] The determination unit 114 makes a determination based on the context state variable and the latest state variable St (S13). Here, the judgment unit 114 uses a judgment learning model, which is a neural network model stored in the storage unit 101, to estimate a judgment from the context state variables and the latest state variables.

[0030] Then, the determination unit 114 stores the determination and the context state variable in association with each other as result information in the storage unit 101, thereby accumulating the result information (S14).

[0031] The evaluation unit 115 uses the result information stored in the storage unit 101 to evaluate the learning state of at least the learning model used by the attention mechanism unit 113. For example, in order to facilitate evaluation, the evaluation unit 115 converts the N-dimensional data obtained by clustering the result information for each judgment into lower-dimensional data (S15). Specifically, the evaluation unit 115 visualizes the clusters for each judgment by converting the N-dimensional data into two-dimensional data using T-sne (T-Eistributed Stochastic Neighbor Embedding).

[0032] Then, the evaluation unit 115 evaluates the learning state by calculating, for example, the distance or similarity between the clusters as an evaluation value (S16). For example, the evaluation unit 115 performs the evaluation by comparing the evaluation value between the clusters with a threshold. Specifically, the evaluation unit 115 determines that the learning is insufficient when the distance between the clusters is less than a predetermined threshold or when the similarity between the clusters is higher than a predetermined threshold. The result of the determination by the evaluation unit 115 may be displayed on the display unit 104, for example.

[0033] As described above, according to the first embodiment, the learning state of a learning model using an attention mechanism can be easily grasped.

[0034] Embodiment 2 FIG. 4 is a block diagram showing a schematic configuration of an information processing device 200 according to the second embodiment. The information processing device 200 includes a storage unit 101 , a communication unit 102 , an input unit 103 , a display unit 104 , and a control unit 210 . The memory unit 101, communication unit 102, input unit 103 and display unit 104 of the information processing device 200 of embodiment 2 are similar to the memory unit 101, communication unit 102, input unit 103 and display unit 104 of the information processing device 100 of embodiment 1.

[0035] The control unit 210 controls the processing in the information processing device 200 . The control unit 210 in the second embodiment performs the same processes as the control unit 110 in the first embodiment, and also performs the following processes. The control unit 210 performs learning of the learning model using additional learning data, which is additional learning data, according to the evaluation result of the learning state.

[0036] The control unit 210 includes a data acquisition unit 111 , a variable extraction unit 112 , an attention mechanism unit 113 , a judgment unit 114 , an evaluation unit 215 , and an additional learning unit 216 . The data acquisition unit 111, the variable extraction unit 112, the attention mechanism unit 113 and the judgment unit 114 of the control unit 210 in embodiment 2 are similar to the data acquisition unit 111, the variable extraction unit 112, the attention mechanism unit 113 and the judgment unit 114 of the control unit 110 in embodiment 1.

[0037] The evaluation unit 215 uses the result information stored in the storage unit 101 to evaluate the learning state of at least the learning model used by the attention mechanism unit 113. Then, the evaluation unit 215 provides the evaluation result to the additional learning unit 216. For example, the evaluation unit 215 generates evaluation information indicating whether or not the learning is sufficient by comparing the evaluation value with a threshold for each combination of two clusters, and provides the evaluation information to the additional learning unit 216.

[0038] The additional learning unit 216 refers to the evaluation information from the evaluation unit 215 and provides additional learning data to the variable extraction unit 112, thereby performing additional learning. Here, when the evaluation by the evaluation unit 215 is lower than a predetermined threshold, the additional learning unit 216 learns at least the attention mechanism learning model using additional learning data that is additional learning data. In the second embodiment, the additional learning unit 216 learns an extraction learning model, a judgment learning model, and an attention mechanism learning model.

[0039] For example, the additional learning unit 216 performs learning by using, as the additional learning data, learning data in which, among a plurality of judgments, a judgment with an evaluation lower than a predetermined threshold is determined to be correct. In other words, the additional learning unit 216 may provide the learning data classified into two clusters determined to be insufficiently learned to the variable extraction unit 112 as the additional learning data. Here, the additional learning data may be acquired from another device, for example, via the communication unit 102, or may be stored in the storage unit 101. The user may instruct, for example, via the input unit 103, from where the additional learning data is to be acquired.

[0040] FIG. 5 is a schematic diagram for explaining the processing in the information processing device 200 according to the second embodiment. The process from S10 to S15 in FIG. 5 is similar to the process from S10 to S15 shown in FIG.

[0041] In the second embodiment, the evaluation unit 215 evaluates the learning state by, for example, calculating the distance or similarity between the clusters as an evaluation value, and generates evaluation information indicating the evaluation result (S26). The evaluation information is information indicating whether or not the learning is sufficient for each combination of two clusters. The generated evaluation information is provided to the additional learning unit 216.

[0042] The additional learning unit 216 refers to the evaluation information and generates additional learning data from the learning data classified into clusters determined to have insufficient learning (S27), and provides the additional learning data to the variable extraction unit 112, thereby performing additional learning.

[0043] As described above, according to the second embodiment, in a learning model using an attention mechanism, it is possible to additionally learn clusters that are insufficiently learned.

[0044] Here, the evaluation unit 215 may use one threshold value to determine whether the learning is sufficient or not, but for example, multiple threshold values ​​can be used to perform risk management of the judgment. Specifically, for judgments between clusters that should not be mistaken, such as "stopping" and "accelerating" a vehicle, the distance must be long or the similarity must be low, so that the risk management of the judgment can be performed by adjusting the threshold value.

[0045] Embodiment 3 FIG. 6 is a block diagram showing a schematic configuration of an information processing device 300 according to the third embodiment. The information processing device 300 includes a storage unit 101 , a communication unit 102 , an input unit 103 , a display unit 104 , and a control unit 310 . The memory unit 101, communication unit 102, input unit 103 and display unit 104 of the information processing device 300 of embodiment 3 are similar to the memory unit 101, communication unit 102, input unit 103 and display unit 104 of the information processing device 100 of embodiment 1.

[0046] The control unit 310 controls the processing in the information processing device 300 . The control unit 310 in the third embodiment performs the same processes as the control unit 110 in the first embodiment, and also performs the following processes. The control unit 310 selects learning data according to the evaluation result of the learning state, and performs learning of the learning model using the selected learning data.

[0047] The control unit 310 includes a data acquisition unit 111 , a variable extraction unit 112 , an attention mechanism unit 113 , a judgment unit 114 , an evaluation unit 315 , a learning data selection unit 317 , and a learning unit 318 . The data acquisition unit 111, the variable extraction unit 112, the attention mechanism unit 113 and the judgment unit 114 of the control unit 310 in embodiment 3 are similar to the data acquisition unit 111, the variable extraction unit 112, the attention mechanism unit 113 and the judgment unit 114 of the control unit 110 in embodiment 1.

[0048] The evaluation unit 315 evaluates the learning state of at least the learning model used by the attention mechanism unit 113, using the result information stored in the storage unit 101, as in the first embodiment. In the third embodiment, the evaluation unit 315 provides the training data selection unit 317 with evaluation value information indicating the evaluation value for each combination of two clusters.

[0049] The learning data selection unit 317 refers to the evaluation value information from the evaluation unit 315 and selects learning data for learning at least the attention mechanism learning model. Here, the learning data selection unit 317 performs selection such that the lower the evaluation corresponding to one judgment, the greater the number of learning data that regards that one judgment as the correct answer. In other words, the learning data selection unit 317 selects learning data such that the lower the evaluation by the evaluation value indicated by the evaluation value information, in other words, the shorter the distance, or the higher the similarity, the greater the number of learning data classified into that cluster. Note that the learning data may be stored in the storage unit 101 or in another device. When the learning data is stored in another device, the learning data selection unit 317 may access the other device via the communication unit 102 to select the learning data.

[0050] The learning unit 318 uses the learning data selected by the learning data selection unit 317 to learn at least the attention mechanism learning model. For example, the learning unit 318 performs learning by providing the learning data selected by the learning data selection unit 317 to the variable extraction unit 112.

[0051] FIG. 7 is a schematic diagram for explaining the processing in the information processing device 300 according to the third embodiment. FIG. 7 shows a process for learning using learning data in the information processing device 300. As a premise, the learning data selection unit 317 provides the learning data selected without referring to the evaluation value information as initial learning data to the learning unit 318. The learning unit 318 performs initial learning by providing the initial learning data to the variable extraction unit 112, and selects learning data according to the evaluation result in the initial learning.

[0052] The process from S11 to S15 in FIG. 7 is similar to the process from S11 to S15 shown in FIG.

[0053] In the third embodiment, the evaluation unit 315 evaluates the learning state by, for example, calculating the distance or similarity between the clusters as an evaluation value, and generates evaluation value information indicating the evaluation value for each combination of two clusters (S36). The generated evaluation value information is provided to the learning data selection unit 317.

[0054] The learning data selection unit 317 refers to the evaluation value information and selects learning data such that the lower the evaluation based on the evaluation value indicated by the evaluation value information, the greater the number of learning data classified into that cluster (S37).Then, the learning data selection unit 317 provides the selected learning data to the learning unit 318.

[0055] The learning unit 318 performs learning by providing the learning data selected by the learning data selection unit 317 to the variable extraction unit 112 (S38).

[0056] As described above, according to the third embodiment, when learning a learning model using an attention mechanism, learning can be performed efficiently by selecting learning data that should be learned with emphasis.

[0057] The learning data selection unit 317 selects learning data such that the lower the evaluation based on the evaluation value indicated by the evaluation value information, the more learning data is classified into that cluster, but the third embodiment is not limited to this example. For example, for a judgment between clusters that should not be mistaken, such as "stopping" and "accelerating" a vehicle, the learning data selection unit 317 can select such that the number of learning data of such clusters increases. Specifically, the learning data selection unit 317 can add or multiply a weight value that lowers the evaluation value of a cluster that should be learned with priority, so that the number of selected learning data increases. For example, such a setting may be performed by a user via the input unit 103.

[0058] Embodiment 4 FIG. 8 is a block diagram showing a schematic configuration of an information processing device 400 according to the fourth embodiment. The information processing device 400 includes a storage unit 101 , a communication unit 102 , an input unit 103 , a display unit 104 , and a control unit 410 . The memory unit 101, communication unit 102, input unit 103 and display unit 104 of the information processing device 400 of embodiment 4 are similar to the memory unit 101, communication unit 102, input unit 103 and display unit 104 of the information processing device 100 of embodiment 1.

[0059] The control unit 410 controls the processing in the information processing device 400 . The control unit 410 in the fourth embodiment performs the same processes as the control unit 110 in the first embodiment, and also performs the following processes. The control unit 410 judges whether or not to continue the learning based on the evaluation result of the learning state, and if it is judged that the learning should be continued, the learning is continued, and if it is judged that the learning should not be continued, the learning is terminated.

[0060] The control unit 410 includes a data acquisition unit 111 , a variable extraction unit 112 , a warning mechanism unit 113 , a judgment unit 114 , an evaluation unit 215 , a learning unit 418 , and a learning continuation judgment unit 419 . The data acquisition unit 111, the variable extraction unit 112, the attention mechanism unit 113 and the judgment unit 114 of the control unit 410 in embodiment 4 are similar to the data acquisition unit 111, the variable extraction unit 112, the attention mechanism unit 113 and the judgment unit 114 of the control unit 110 in embodiment 1. Moreover, the evaluation unit 215 in the fourth embodiment is the same as the evaluation unit 215 in the second embodiment. However, in the fourth embodiment, the evaluation unit 215 provides evaluation information to the learning continuation determination unit 419.

[0061] The learning continuation decision unit 419 refers to the evaluation information from the evaluation unit 215 and decides whether or not to continue learning of at least the attention mechanism learning model. For example, the learning continuation judgment unit 419 judges to continue learning when all or some of the evaluations based on the evaluation values ​​indicated in the evaluation information are lower than a predetermined threshold, in other words, when the distance is shorter than a predetermined threshold or the similarity is higher than a predetermined threshold.

[0062] The part of the evaluations may be a predetermined number of evaluations or may be evaluations of a predetermined cluster. For example, when all evaluations of important clusters that cannot be allowed to be incorrect are equal to or greater than a threshold, the learning continuation determination unit 419 may determine not to continue the learning.

[0063] When the learning continuation determination unit 419 determines to continue learning, the learning unit 418 performs learning by providing the learning data to the variable extraction unit 112. On the other hand, when the learning continuation determination unit 419 determines not to continue learning, the learning unit 418 does not provide the learning data to the variable extraction unit 112 and ends learning. The learning data may be stored in the storage unit 101 or in another device. When the learning data is stored in another device, the learning unit 418 may access the other device via the communication unit 102 to acquire the learning data.

[0064] FIG. 9 is a schematic diagram for explaining the processing in the information processing device 400 according to the fourth embodiment. FIG. 9 shows a process for learning using learning data in the information processing device 400. As a premise, the learning unit 418 performs initial learning by providing learning data to the variable extraction unit 112 as initial learning data, and determines whether to continue the learning depending on the evaluation result of the initial learning.

[0065] The processes from S11 to S15 in FIG. 9 are similar to the processes from S11 to S15 shown in FIG.

[0066] In the fourth embodiment, the evaluation unit 215 evaluates the learning state by, for example, calculating the distance or similarity between the clusters as an evaluation value, and generates evaluation information indicating the evaluation result (S46). The evaluation information is information indicating whether or not the learning is sufficient for each combination of two clusters. The generated evaluation information is provided to the learning continuation judgment unit 419.

[0067] The learning continuation decision unit 419 refers to the evaluation information from the evaluation unit 215 and decides whether or not to continue the learning (S47).

[0068] Then, when the learning continuation decision unit 419 decides to continue the learning, the learning unit 418 performs learning by providing the learning data to the variable extraction unit 112 (S48).

[0069] As described above, according to the fourth embodiment, when learning a learning model using an attention mechanism, if the learning is sufficiently performed, the learning can be terminated. Therefore, the learning can be performed efficiently.

[0070] As in the second embodiment, the evaluation unit 215 may use one threshold value to determine whether learning is sufficient, but for example, multiple threshold values ​​can be used to manage the risk of judgment. Specifically, for judgments between clusters that should not be mistaken, such as "stopping" and "accelerating" a vehicle, the distance must be long or the similarity must be small, so that the risk of judgment can be managed by adjusting the threshold value. [Explanation of symbols]

[0071] 100,200,300,400 Information processing device, 101 Memory unit, 102 Communication unit, 103 Input unit, 104 Display unit, 110,210,310,410 Control unit, 111 Data acquisition unit, 112 Variable extraction unit, 113 Attention mechanism unit, 114 Judgment unit, 115,215,315 Evaluation unit, 216 Additional learning unit, 317 Learning data selection unit, 318,418 Learning unit, 419 Learning continuation judgment unit.

Claims

1. an attention mechanism unit that calculates a context variable by weighting and adding multiple time-series variables using an attention mechanism learning model that is a learning model of an attention mechanism; a judgment unit that estimates one judgment included in the plurality of judgments based on a reliability of the plurality of judgments calculated from the context variable and a latest variable included in the plurality of variables; a storage unit that stores result information in which the context variable and the one judgment are associated with each other; and an evaluation unit that evaluates at least the learning state of the attention mechanism learning model based on the result information. An information processing device comprising:

2. The judgment unit estimates the one judgment using a judgment learning model which is a learning model for estimating the one judgment from the context variable, The evaluation unit performs the evaluation of the judgment learning model and the attention mechanism learning model.

2. The information processing device according to claim 1,

3. Further comprising a variable extraction unit for extracting the variables from the input data.

3. The information processing apparatus according to claim 2,

4. The variable extraction unit extracts the variables using an extraction learning model that is a learning model for extracting the variables from the input data; The evaluation unit performs the evaluation of the extraction learning model, the judgment learning model, and the attention mechanism learning model.

4. The information processing apparatus according to claim 3,

5. Further comprising a variable extraction unit for extracting the variables from the input data.

2. The information processing device according to claim 1,

6. The variable extraction unit extracts the variables using an extraction learning model that is a learning model for extracting the variables from the input data; The evaluation unit performs the evaluation of the extraction learning model and the attention mechanism learning model.

6. The information processing device according to claim 5,

7. The evaluation unit performs clustering for each of the plurality of judgments to identify a plurality of clusters, and performs the evaluation based on a distance or a similarity between the plurality of clusters. The information processing device according to claim 1 ,

8. Further comprising an additional learning unit that learns at least the attention mechanism learning model using additional learning data that is additional learning data when the evaluation is lower than a predetermined threshold value. The information processing device according to claim 1 ,

9. the additional learning unit uses, as the additional learning data, learning data in which a judgment in which the evaluation is lower than the predetermined threshold is determined to be a correct answer among the plurality of judgments. The information processing device according to claim 8 .

10. a learning data selection unit that selects learning data for training at least the attention mechanism learning model according to the evaluation; A learning unit that uses the selected learning data to learn at least the attention mechanism learning model. The information processing device according to claim 1 ,

11. The learning data selection unit performs the selection such that the lower the evaluation in one judgment, the greater the number of the learning data that determines the one judgment as a correct answer. The information processing device according to claim 10 .

12. a learning continuation determination unit that determines whether or not to continue learning at least the attention mechanism learning model according to the evaluation; and a learning unit that, when it is determined that the learning is to be continued, continues the learning using at least learning data for learning the attention mechanism learning model, and, when it is determined that the learning is not to be continued, ends the learning. The information processing device according to claim 1 ,

13. The learning continuation determination unit determines to continue the learning when the evaluation of all or a part of the plurality of judgments is lower than a predetermined threshold value. The information processing device according to claim 12 .

14. Computer, An attention mechanism unit that calculates a context variable by weighting and adding multiple time-series variables using an attention mechanism learning model that is a learning model of an attention mechanism; a judgment unit that estimates one judgment included in the plurality of judgments based on reliability of the plurality of judgments calculated from the context variable and a latest variable included in the plurality of variables; a storage unit that stores result information in which the context variable and the one judgment are associated with each other; and and functioning as an evaluation unit that evaluates at least the learning state of the attention mechanism learning model based on the result information. A program characterized by.

15. An attention mechanism unit calculates a context variable by weighting and adding a plurality of time-series variables using an attention mechanism learning model, which is a learning model of an attention mechanism; a judgment unit that estimates one judgment included in the plurality of judgments based on a reliability of the plurality of judgments calculated from the context variable and a latest variable included in the plurality of variables; a storage unit stores result information in which the context variable and the one judgment are associated with each other; An evaluation unit evaluates at least the learning state of the attention mechanism learning model based on the result information. An information processing method comprising:

Images