Information processing device, information processing method, and recording medium

Abstract

An information processing device (10) comprises: an estimation means (100) for estimating, in response to input of first sequence data, second sequence data corresponding to the first sequence data; an acquisition means (50) for acquiring a set of the first sequence data and the second sequence data; a quantization means (60) for quantizing a second element included in the second sequence data; an encoding means (70) for encoding the quantized second element; and a training means (300) for respectively assigning a value for each digit of the encoded second element to a plurality of classification means (101) included in the estimation means, and training each of the plurality of classification means by using a first element included in the first sequence data and the assigned second element. Such an information processing device makes it possible to appropriately estimate a real value from sequence data.

Description

Information processing device, information processing method, and recording medium

[0001] The present disclosure relates to the technical fields of an information processing device, an information processing method, and a recording medium.

[0002] Known examples of this type of device include one that classifies sequential data using likelihood ratios. For example, Patent Document 1 discloses a device that sequentially acquires and analyzes multiple elements included in the sequential data to classify the sequential data into one of multiple predetermined classes.

[0003] International Publication No. 2020 / 194497

[0004] This disclosure aims to improve upon the related art discussed above.

[0005] One aspect of the information processing device disclosed herein comprises an estimation means for estimating second sequence data corresponding to first sequence data by inputting the first sequence data, an acquisition means for acquiring a pair of the first sequence data and the second sequence data, a quantization means for quantizing a second element included in the second sequence data, an encoding means for encoding the quantized second element, and a learning means for assigning the value of each digit of the encoded second element to a plurality of classification means included in the estimation means, and for training each of the plurality of classification means using the first element included in the first sequence data and the assigned second element.

[0006] Another aspect of the information processing device disclosed herein comprises a data acquisition means for acquiring first series data, a plurality of classification means for inputting a plurality of first elements included in the first series data and estimating the values ​​of each digit of an encoded second element corresponding to the first elements, and a decoding means for obtaining second series data corresponding to the first series data by combining and decoding the values ​​of each digit of the second element estimated by each of the plurality of classification means.

[0007] One aspect of the information processing method disclosed herein is an information processing method for training an estimation means for estimating second sequence data corresponding to first sequence data by inputting the first sequence data, the information processing method comprising: acquiring a pair of the first sequence data and the second sequence data; quantizing a second element included in the second sequence data; encoding the quantized second element; assigning the values ​​of each digit of the encoded second element to a plurality of classification means possessed by the estimation means; and training each of the plurality of classification means using the first element included in the first sequence data and the assigned second element.

[0008] Another aspect of the information processing method of this disclosure involves obtaining first sequence data, inputting a plurality of first elements contained in the first sequence data, estimating the values ​​of each digit of an encoded second element corresponding to the first elements, and decoding the estimated values ​​of each digit of the second elements together to obtain second sequence data corresponding to the first sequence data.

[0009] One aspect of the recording medium of this disclosure is an information processing method for training an estimation means for estimating second sequence data corresponding to first sequence data by inputting the first sequence data, wherein a computer program is recorded on the recording medium to cause a computer to execute the information processing method, which includes acquiring a pair of the first sequence data and the second sequence data, quantizing a second element included in the second sequence data, encoding the quantized second element, assigning the value of each digit of the encoded second element to a plurality of classification means possessed by the estimation means, and training each of the plurality of classification means using the first element included in the first sequence data and the assigned second element.

[0010] Another aspect of the recording medium of this disclosure is a recording medium having a computer program recorded thereon that causes a computer to execute an information processing method, which involves acquiring first sequence data, inputting a plurality of first elements contained in the first sequence data, estimating the values ​​of each digit of an encoded second element corresponding to the first elements, and combining and decoding the values ​​of each digit of the estimated second elements to obtain second sequence data corresponding to the first sequence data.

[0011] 1 is a block diagram showing a hardware configuration of a first information processing device; FIG. 2 is a block diagram showing a functional configuration of the first information processing device; FIG. 3 is a graph showing an example of quantization and encoding processing in the first information processing device; FIG. 4 is a graph showing an example of estimated values ​​estimated in the first information processing device; FIG. 5 is a graph showing a method of classifying estimated values ​​in the first information processing device; FIG. 6 is a flowchart showing the flow of operation of the first information processing device; FIG. 7 is a block diagram showing the functional configuration of a second information processing device; and FIG. 8 is a flowchart showing the flow of operation of the second information processing device.

[0012] Hereinafter, embodiments of an information processing device, an information processing method, and a recording medium will be described with reference to the drawings.

[0013] First Embodiment A first embodiment will be described with reference to FIGS. 1 to 6. FIG.

[0014] (Hardware Configuration) First, the hardware configuration of the first information processing apparatus will be described with reference to Fig. 1. Fig. 1 is a block diagram showing the hardware configuration of the first information processing apparatus.

[0015] 1, a first information processing device 10 includes a processor 11, a RAM (Random Access Memory) 12, a ROM (Read Only Memory) 13, and a storage device 14. The information processing device 10 may further include an input device 15 and an output device 16. The processor 11, RAM 12, ROM 13, storage device 14, input device 15, and output device 16 are connected to each other via a data bus 17.

[0016] The processor 11 loads a computer program. For example, the processor 11 is configured to load a computer program stored in at least one of the RAM 12, the ROM 13, and the storage device 14. Alternatively, the processor 11 may load a computer program stored in a computer-readable storage medium using a storage medium reading device (not shown). The processor 11 may acquire (i.e., load) the computer program from a device (not shown) located outside the information processing device 10 via a network interface. The processor 11 controls the RAM 12, the storage device 14, the input device 15, and the output device 16 by executing the loaded computer program. In particular, in this embodiment, when the processor 11 executes the loaded computer program, functional blocks that perform various processes related to likelihood ratios are realized within the processor 11. In other words, the processor 11 may function as a controller that executes each control in the information processing device 10.

[0017] The processor 11 may be configured as, for example, a central processing unit (CPU), a graphics processing unit (GPU), a field-programmable gate array (FPGA), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), or a quantum processor. The processor 11 may be configured as one of these, or may be configured to use multiple processors in parallel.

[0018] The RAM 12 temporarily stores computer programs executed by the processor 11. The RAM 12 temporarily stores data that the processor 11 temporarily uses while it is executing the computer programs. The RAM 12 may be, for example, a dynamic random access memory (D-RAM) or a static random access memory (SRAM). Alternatively, other types of volatile memory may be used instead of the RAM 12.

[0019] The ROM 13 stores computer programs executed by the processor 11. The ROM 13 may also store fixed data. The ROM 13 may be, for example, a programmable read-only memory (PROM) or an erasable read-only memory (EPROM). Alternatively, other types of non-volatile memory may be used instead of the ROM 13.

[0020] The storage device 14 stores data that is to be saved long-term by the information processing device 10. The storage device 14 may operate as a temporary storage device for the processor 11. The storage device 14 may include, for example, at least one of a hard disk device, a magneto-optical disk device, an SSD (Solid State Drive), and a disk array device.

[0021] The input device 15 is a device that receives input instructions from a user of the information processing device 10. The input device 15 may include, for example, at least one of a keyboard, a mouse, and a touch panel. The input device 15 may be configured as a mobile terminal such as a smartphone or a tablet. The input device 15 may also be, for example, a device that includes a microphone and is capable of voice input.

[0022] The output device 16 is a device that outputs information related to the information processing device 10 to the outside. For example, the output device 16 may be a display device (e.g., a display) that can display information related to the information processing device 10. The output device 16 may also be a speaker or the like that can output information related to the information processing device 10 as audio. The output device 16 may be configured as a mobile terminal such as a smartphone or a tablet.

[0023] 1 shows an example of an information processing device 10 including multiple devices, but all or some of the functions may be realized by a single device. Such an information processing device may be configured to include only the above-mentioned processor 11, RAM 12, and ROM 13, and the other components (i.e., the storage device 14, input device 15, output device 16, etc.) may be provided by an external device connected to the information processing device 10. Furthermore, some of the calculation functions of the information processing device 10 may be realized by an external device (e.g., an external server, a cloud, etc.).

[0024] (Functional Configuration) Next, the functional configuration of the first information processing device 10 will be described with reference to Fig. 2. Fig. 2 is a block diagram showing the functional configuration of the first information processing device.

[0025] 2, the first information processing device 10 is configured to include, as components for realizing its functions, an acquisition unit 50, a quantization unit 60, an encoding unit 70, an estimation unit 100, and a learning unit 300. Each of the acquisition unit 50, the quantization unit 60, the encoding unit 70, the estimation unit 100, and the learning unit 300 may be a processing block realized by, for example, the above-mentioned processor 11 (see FIG. 1).

[0026] The acquisition unit 50 is configured to be able to acquire sequence data. Sequence data here refers to data including multiple elements arranged in a predetermined order, such as time-series data. More specific examples of sequence data include, but are not limited to, video data, audio data, or subdivided image data. The acquisition unit 50 may acquire data directly from any data acquisition device (e.g., a camera, a microphone, etc.), or may read data that has been acquired in advance by a data acquisition device and stored in storage, etc. When acquiring data from a camera, the acquisition unit 50 may be configured to acquire data from each of multiple cameras.

[0027] The acquisition unit 50 of the first information processing device 10 is particularly configured to acquire a set of first sequence data and second sequence data corresponding to the first sequence data. The second sequence data is data including a second element, which is a real value, and is a value corresponding to an estimated value estimated by inputting the first element included in the first sequence data to the estimation unit 100. In other words, the second sequence data is acquired as correct answer data in the learning of the estimation unit 100. The combination of the first sequence data and the second sequence data is not particularly limited as long as there is a correspondence between them. For example, if the first sequence data is an image of a pedestrian, the second sequence data may be the facial orientation of the pedestrian. Alternatively, if the first sequence data is a satellite image, the second sequence data may be the amount of precipitation at the corresponding time. The set of the first sequence data and the second sequence data may be generated by a prior annotation operation.

[0028] The quantization unit 60 is configured to be able to quantize the second element included in the second sequence data out of the sequence data acquired by the acquisition unit 50. That is, the quantization unit 60 may be configured to be able to perform a process of removing significant digits from the second element, which is a real value. The specific operation of the quantization unit 60 will be described in detail later.

[0029] The encoding unit 70 is configured to be able to encode the second element quantized by the quantization unit 60. The encoding unit 70 may be configured to be able to execute a process of encoding the second element based on a predetermined rule. A more specific operation of the encoding unit 70 will be described in detail later.

[0030] The estimation unit 100 is configured to be able to estimate second sequence data from the first sequence data acquired by the acquisition unit 50. More specifically, the estimation unit 100 is configured to be able to estimate multiple second elements included in the second sequence data from multiple first elements included in the first sequence data. The estimation unit 100 according to this embodiment estimates the second elements using multiple classification units 101a to 101k. The number (k) of classification units 101 corresponds to the number of digits of the second elements to be estimated. Specifically, each classification unit 101 estimates a value corresponding to each digit of the second element. For example, classification unit 101a estimates the first digit value, classification unit 101b estimates the second digit value, ..., classification unit 101k estimates the kth digit value. This number of digits k corresponds to the number of digits of the second elements encoded by the encoding unit 70.

[0031] Each of the multiple classification units 101 may estimate the value of each digit using a likelihood ratio. The classification unit 101 selects at least one class to which each piece of data belongs from multiple classes that are classification candidates. For example, the classification unit 101 may perform binary classification of "1" or "0" using the likelihood ratio. Note that the "likelihood ratio" here is an index indicating the likelihood of the class to which each piece of data belongs. The classification unit 101 may be configured to be able to calculate the likelihood ratio based on the relationship between at least two consecutive elements among multiple first elements included in the first sequence data. A specific classification method using the likelihood ratio will be described in detail later.

[0032] The learning unit 300 is configured to train each of the multiple classifiers 101 included in the estimation unit 100. Specifically, the learning unit 300 trains each of the multiple classifiers 101 using a first element included in the first sequence data and a second element, which is the correct answer data for the first element. The second element used as the correct answer data is the second element quantized by the quantization unit 60 and coded by the coding unit 70. The quantized and coded second elements are assigned to the corresponding classifiers 101. For example, the classifier 101a may be trained using the first digit value of the coded second element. Similarly, the classifier 101k may be trained using the kth digit value of the coded second element. The learning method is not particularly limited, and the learning unit 300 may train the classifiers 101 using, for example, backpropagation. Since existing technologies (see, for example, References 1 and 2 below) can be appropriately adopted as the learning method, detailed description thereof will be omitted here.

[0033] [Reference 1]: International Publication No. 2021 / 229663 [Reference 2]: International Publication No. 2022 / 157973

[0034] (Quantization and Encoding) Next, the operations of the quantization unit 60 and the encoding unit 70 will be described with reference to Fig. 3. Fig. 3 is a graph showing an example of the quantization and encoding process in the first information processing device.

[0035] 3, the second elements (i.e., elements included in the second sequence data), which are real-valued values ​​acquired by the acquisition unit 50, are first quantized by the quantization unit 60. The quantized values ​​become discontinuous values ​​as shown in the figure.

[0036] The quantized values ​​are then coded by the coding unit 70. Here, binary coding (coding using 0 / 1 codes) is used as an example, but other coding methods may be used. When coding into two-valued values, such as binary coding, the classification unit 101 may be configured to be capable of binary classification. Furthermore, if coding is performed so that three or more values ​​are possible, the classification unit 101 may be configured to be capable of multi-class classification of three or more values.

[0037] (Estimation of Real Values) Next, the estimation operation by the estimation unit 100 will be specifically described with reference to Fig. 4. Fig. 4 is a graph showing an example of estimated values ​​estimated by the first information processing apparatus.

[0038] As shown in FIG. 4, the estimation unit 100 calculates the first element x ｔ The second element y corresponding to ｔ For example, at time t １ The first element x １ When the second element y １ Similarly, at time t ２ The first element x ２ When the second element y ２ Output.

[0039] The second element y ｔ is output from each of the classification units 101 as a value corresponding to each digit of the binary code. １ The second element y corresponding to １ is output as "1, 1, 0, 0". ２ The second element y corresponding to ２ is output as "1, 1, 1, 0." Such a four-digit binary code can be estimated using four classification units 101 (i.e., a classification unit 101 corresponding to each digit).

[0040] (Classification by Likelihood Ratio) Next, the classification operation by the classification unit 101 will be specifically described with reference to Fig. 5. Fig. 5 is a graph showing a method for classifying estimated values ​​in the first information processing apparatus.

[0041] As shown in Fig. 5, each digit of the binary code is estimated by a corresponding classifier 101. Specifically, the value of each digit of the coded second element is estimated based on likelihood ratios calculated by the multiple classifiers 101. In the example of classifier A shown in the figure, １ ~t ３ During the time t １ ~t ３The value of the binary code corresponding to time t ４ ~t ７ The likelihood ratio is calculated as the value on the label 0 side during time t ４ ~t ７ The value of the binary code corresponding to is estimated as "0". Similarly, values ​​based on likelihood ratios are estimated in the other classification units B to D, and a four-digit binary code is estimated at each time point based on these estimation results.

[0042] As described above, each of the plurality of classifiers 101 is assigned a first element x ｔ When the first element x is input, the quantized and coded second element is output. ｔ By using a set of the first element and the quantized and coded second element, each of the plurality of classifiers 101 can be trained.

[0043] (Flow of Operation) Next, the flow of operation of the first information processing device 10 will be described with reference to Fig. 6. Fig. 6 is a flowchart showing the flow of operation of the first information processing device.

[0044] 6, when the information processing device 10 starts operating, the acquiring unit 50 first acquires a set of first sequence data X and second sequence data Y (step S101). The first sequence data X is composed of a plurality of first elements x ｔ The second sequence data Y includes a plurality of second elements y ｔ The first sequence data X and the second sequence data Y do not have to be acquired all at once, but may be acquired sequentially in the order of their sequences. For example, the first sequence data X and the second sequence data Y may be acquired sequentially in element units contained therein. Here, if the time at which each element is acquired is t={1, 2, ..., i, ..., T}, the first sequence data X is （１,Ｔ） = {x １ , x ２ , ..., x ｉ , ..., x Ｔ}, the second column data Y is Y （１,Ｔ） = {y １ , y ２ , ..., y ｉ , ..., y Ｔ}.

[0045] Next, the quantization unit 60 calculates the second element y ｔ (Step S102). Then, the encoding unit 70 quantizes the second element y ｔ The encoded second element y is then encoded (step S103). ｔ is y ｔｋ = {y ｔ１ , y ｔ２ , ..., y ｔＫ}.

[0046] Next, the learning unit 300 calculates the label of the kth classifier 101 at time t as the second element y ｔｋ (i.e., the encoded second element y ｔ In other words, each of the multiple classification units 101 receives the encoded second element y ｔ Assign a value to each digit of

[0047] Next, the information processing device 10 determines whether or not there is any unacquired sequence data (i.e., a set of first sequence data X and second sequence data Y) remaining (step S105). If there is any unacquired sequence data remaining (step S105: YES), the time t is set to t+1 (step S106), and the process is repeated from step S101. That is, the first element x ｔ and the second element y ｔ and obtain the second element y ｔ The process of quantizing and encoding is repeated.

[0048] On the other hand, if there is no unacquired sequence data remaining (step S105: NO), the learning unit 300 sets X （１,Ｔ） = {x １ , x ２ , ..., x ｉ , ..., x Ｔ}, and Y （１,Ｔ） = {y １ｋ , y ２ｋ , ..., y ｉｋ , ..., y Ｔｋ} is obtained (step S107). Then, the learning unit 300 uses these learning data to perform learning on each of the multiple classifying units 101 (step S108).

[0049] (Technical Effects) Next, technical effects obtained by the first information processing device 10 will be described.

[0050] 1 to 6 , in the first information processing device, of the first sequence data X and second sequence data Y acquired as training data, the second sequence data Y corresponding to the correct answer data is quantized and coded before being used for training. In this way, it is possible to appropriately train the estimation unit 100 (specifically, the plurality of classifiers 101) that estimates real values.

[0051] It has been found that the method of regressing quantized and coded values ​​by classification, as in this embodiment, is more accurate than simple regression methods (e.g., regression methods using squared error). One reason for this is that, depending on the binary encoding format, some classification errors can be corrected through error correction capabilities. Furthermore, as described in this embodiment, the regression accuracy can be further improved by calculating the likelihood ratio taking into account at least two consecutive data points.

[0052] Second Embodiment A second embodiment will be described with reference to Figures 7 and 8. The second embodiment differs from the first embodiment described above only in some configurations and operations, and other parts may be the same as the first embodiment. Therefore, the following will describe in detail the parts that differ from the first embodiment, and will omit a description of other overlapping parts as appropriate.

[0053] (Functional Configuration) First, the functional configuration of the second information processing device 10 will be described with reference to Fig. 7. Fig. 7 is a block diagram showing the functional configuration of the second information processing device. Note that in Fig. 7, the same elements as those shown in Fig. 2 are denoted by the same reference numerals.

[0054] 7, the second information processing device 10 is a device for performing estimation using the estimation unit 100 trained in the first information processing device (see FIG. 2) (i.e., estimating second sequence data from first sequence data). The second information processing device 10 is configured to include an acquisition unit 50, a decoding unit 80, and an estimation unit 100 as components for realizing its functions. That is, while the second information processing device 10 includes the decoding unit 80 that was not included in the configuration described in the first embodiment (see FIG. 2), it does not include the quantization unit 60, encoding unit 70, and learning unit 300 used for learning. The decoding unit 80 may be a processing block realized by, for example, the above-mentioned processor 11 (see FIG. 1).

[0055] The decoding unit 80 decodes the second element y output from the estimation unit 100. ｔｋ That is, the decoding unit 80 is configured to be able to decode the decoded second element y ｔｋ The second element y ｔ The decoding process performed by the decoding unit 80 may be a decoding process corresponding to the encoding process performed by the encoding unit 80 (see FIG. 2) described in the first embodiment.

[0056] The acquisition unit 50 in the second information processing device 10 may be configured to acquire only the first sequence data. That is, since it is assumed that the learning operation of the estimation unit 100 described in the first embodiment has already been completed, there is no need to acquire a set of the first sequence data and the second sequence data, which are learning data, as in the acquisition unit in the first information processing device 10 (see FIG. 2). However, learning may be performed again while the device is in operation. In this case, the second information processing device 10 may be configured to include a quantization unit 60, an encoding unit 70, and a learning unit 300, like the first information processing device 10.

[0057] (Flow of Operation) Next, the flow of operation of the second information processing device 10 will be described with reference to Fig. 8. Fig. 8 is a flowchart showing the flow of operation of the second information processing device.

[0058] 8, when the operation of the second information processing device 10 is started, the acquisition unit 50 first acquires the first sequence data X (step S201). The first sequence data X is composed of a plurality of first elements x ｔ The first sequence data X does not have to be acquired all at once, but may be acquired sequentially in the sequence order. For example, the first sequence data X may be acquired sequentially in element units contained therein. Here, if the time at which each element is acquired is t={1, 2, ..., i, ..., T}, the first sequence data X is （１,Ｔ） = {x １ , x ２ , ..., x ｉ , ..., x Ｔ}.

[0059] Next, the first element x included in the first sequence data X acquired by the acquisition unit ｔ is input to the estimation unit 100. As a result, the second element y ｔｋ (Step S202). The second element y ｔｋ is in an encoded state (for example, binary code). Then, the decoding unit 80 decodes the second element y ｔｋ (Step S203). As a result, the second element y ｔ get.

[0060] Next, the information processing device 10 extracts the unacquired first sequence data X (i.e., the unacquired first element x ｔ If there is any unacquired first sequence data X remaining (step S204: YES), the time t is set to t+1 (step S205), and the process is repeated from step S201. That is, the first element x ｔ to the second element y ｔｋ and decode it to obtain the second element y ｔ Repeat the process to obtain

[0061] On the other hand, if there is no unacquired first sequence data X remaining (step S204: NO), the information processing device outputs second sequence data Y', which is the estimation result (step S206).１ , y′ ２ , ..., y' Ｔ}.

[0062] (Technical Effects) Next, technical effects obtained by the second information processing device 10 will be described.

[0063] As described with reference to FIGS. 7 and 8, the second information processing apparatus 10 uses a plurality of classification units 101 to classify the encoded second element y ｔｋ and decoding the digits to obtain the second sequence data Y', which is a real value. In this way, the second sequence data Y', which is a real value, can be appropriately estimated from the first sequence data X. In other words, it is possible to solve a regression problem with high accuracy using the classification unit 101.

[0064] The scope of each embodiment also includes a processing method in which a program that operates the configuration of each embodiment to realize the functions of the above-described embodiments is recorded on a recording medium, the program recorded on the recording medium is read as code, and the program is executed on a computer. In other words, a computer-readable recording medium is also included in the scope of each embodiment. Furthermore, each embodiment includes not only a recording medium on which the above-described program is recorded, but also the program itself.

[0065] Examples of recording media that can be used include floppy disks, hard disks, optical disks, magneto-optical disks, CD-ROMs, magnetic tapes, non-volatile memory cards, and ROMs. Furthermore, the scope of each embodiment is not limited to programs that execute processes by themselves, but also includes programs that execute processes by operating on an OS in conjunction with other software or expansion board functions. Furthermore, the program itself may be stored on a server, and part or all of the program may be downloadable from the server to a user terminal. The program may be provided to the user in, for example, a SaaS (Software as a Service) format.

[0066] <Supplementary Notes> The above-described embodiment may be further described as in the following supplementary notes, but is not limited to the following.

[0067] (Supplementary Note 1) The information processing device described in Supplementary Note 1 is an information processing device comprising: an estimation means for estimating second sequence data corresponding to first sequence data by inputting the first sequence data; an acquisition means for acquiring a pair of the first sequence data and the second sequence data; a quantization means for quantizing a second element included in the second sequence data; an encoding means for encoding the quantized second element; and a learning means for assigning the value of each digit of the encoded second element to a plurality of classification means included in the estimation means, and for training each of the plurality of classification means using the first element included in the first sequence data and the assigned second element.

[0068] (Supplementary Note 2) The information processing device described in Supplementary Note 2 is the information processing device described in Supplementary Note 1, in which the plurality of classification means are trained as a model that calculates likelihood ratios for the values ​​of each digit of the encoded second element by inputting the first elements that are adjacent to each other.

[0069] (Supplementary Note 3) The information processing device described in Supplementary Note 3 is an information processing device comprising: a data acquisition means for acquiring first series data; a plurality of classification means for inputting a plurality of first elements included in the first series data and estimating the value of each digit of an encoded second element corresponding to the first elements; and a decoding means for obtaining second series data corresponding to the first series data by combining and decoding the values ​​of each digit of the second element estimated by each of the plurality of classification means.

[0070] (Supplementary Note 4) The information processing method described in Supplementary Note 4 is an information processing method for training an estimation means for estimating second sequence data corresponding to first sequence data by inputting the first sequence data, the information processing method comprising: acquiring a pair of the first sequence data and the second sequence data; quantizing a second element included in the second sequence data; encoding the quantized second element; assigning values ​​of each digit of the encoded second element to a plurality of classification means included in the estimation means; and training each of the plurality of classification means using the first element included in the first sequence data and the assigned second element.

[0071] (Supplementary Note 5) The information processing method described in Supplementary Note 5 is an information processing method that acquires first sequence data, inputs a plurality of first elements included in the first sequence data, estimates the value of each digit of an encoded second element corresponding to the first elements, and decodes the estimated values ​​of each digit of the second element together to obtain second sequence data corresponding to the first sequence data.

[0072] (Supplementary Note 6) The recording medium described in Supplementary Note 6 is a recording medium having recorded thereon a computer program that causes a computer to execute an information processing method for training an estimation means that receives first sequence data and estimates second sequence data corresponding to the first sequence data, the information processing method comprising: acquiring a pair of the first sequence data and the second sequence data; quantizing a second element included in the second sequence data; encoding the quantized second element; assigning values ​​of each digit of the encoded second element to a plurality of classification means included in the estimation means; and training each of the plurality of classification means using the first element included in the first sequence data and the assigned second element.

[0073] (Supplementary Note 7) The recording medium described in Supplementary Note 7 is a recording medium having recorded thereon a computer program that causes a computer to execute an information processing method, which involves acquiring first sequence data, inputting a plurality of first elements included in the first sequence data, estimating the values ​​of each digit of encoded second elements corresponding to the first elements, and combining and decoding the values ​​of each digit of the estimated second elements to obtain second sequence data corresponding to the first sequence data.

[0074] (Supplementary Note 8) The computer program described in Supplementary Note 8 is an information processing method for training an estimation means for estimating second sequence data corresponding to first sequence data by inputting the first sequence data, the computer program causing a computer to execute the information processing method, which includes acquiring a pair of the first sequence data and the second sequence data, quantizing a second element included in the second sequence data, encoding the quantized second element, assigning values ​​of each digit of the encoded second element to a plurality of classification means included in the estimation means, and training each of the plurality of classification means using the first element included in the first sequence data and the assigned second element.

[0075] (Supplementary Note 9) The computer program described in Supplementary Note 9 is a computer program that causes a computer to execute an information processing method, which acquires first sequence data, inputs a plurality of first elements included in the first sequence data, estimates the value of each digit of an encoded second element corresponding to the first elements, and decodes the estimated values ​​of each digit of the second elements together to obtain second sequence data corresponding to the first sequence data.

[0076] This disclosure may be modified as appropriate within the scope that does not contradict the gist or idea of ​​the invention that can be read from the claims and the entire specification, and information processing devices, information processing methods, and recording media that involve such modifications are also included in the technical idea of ​​this disclosure.

[0077] REFERENCE SIGNS LIST 10 Information processing device 11 Processor 50 Acquisition unit 60 Quantization unit 70 Encoding unit 80 Decoding unit 100 Estimation unit 101 Classification unit 300 Learning unit

Claims

1. an estimation means for estimating second sequence data corresponding to the first sequence data by receiving the first sequence data; an acquisition means for acquiring a set of the first sequence data and the second sequence data; quantization means for quantizing a second element included in the second sequence data; encoding means for encoding the quantized second component; a learning means for assigning the value of each digit of the encoded second element to each of a plurality of classification means included in the estimation means, and for training each of the plurality of classification means using the first element included in the first sequence data and the assigned second element; A learning device comprising:

2. the plurality of classification means are trained as a model that calculates likelihood ratios for values ​​of each digit of the encoded second element by inputting the first elements adjacent to each other; The learning device according to claim 1 .

3. a data acquisition means for acquiring first sequence data; a plurality of classification means for inputting a plurality of first elements included in the first sequence data and estimating values ​​of each digit obtained by encoding a second element corresponding to the first elements; a decoding means for decoding the values ​​of each digit of the second element estimated by each of the plurality of classification means to obtain second sequence data corresponding to the first sequence data; An estimation device comprising:

4. 1. An information processing method for learning an estimation means for estimating second sequence data corresponding to first sequence data by inputting the first sequence data, comprising: acquiring a set of the first sequence data and the second sequence data; quantizing a second element included in the second sequence data; encoding the quantized second component; assigning the value of each digit of the encoded second element to each of a plurality of classification means included in the estimation means, and training each of the plurality of classification means using the first element included in the first sequence data and the assigned second element; Information processing methods.

5. Acquire the first series of data; by inputting a plurality of first elements included in the first sequence data, estimating values ​​of each digit obtained by encoding second elements corresponding to the first elements; obtaining second sequence data corresponding to the first sequence data by combining and decoding the estimated values ​​of each digit of the second element; Information processing methods.

6. 1. An information processing method for learning an estimation means for estimating second sequence data corresponding to first sequence data by inputting the first sequence data, comprising: acquiring a set of the first sequence data and the second sequence data; quantizing a second element included in the second sequence data; encoding the quantized second component; assigning the value of each digit of the encoded second element to each of a plurality of classification means included in the estimation means, and training each of the plurality of classification means using the first element included in the first sequence data and the assigned second element; A computer program that causes a computer to execute an information processing method.

7. Acquire the first series of data; by inputting a plurality of first elements included in the first sequence data, estimating values ​​of each digit obtained by encoding second elements corresponding to the first elements; obtaining second sequence data corresponding to the first sequence data by combining and decoding the estimated values ​​of each digit of the second element; A computer program that causes a computer to execute an information processing method.