Data processing device, data processing system and data processing method

Abstract

To provide a data processing apparatus, a data processing system, and a data processing method which allow for data processing with high accuracy.SOLUTION: In accordance with an embodiment, the data processing apparatus includes a processing unit. In a first operation, the processing unit can generate a first machine learning model on the basis of first generated data based on first acquired data and first other data. The first other data includes a first other feature quantity matrix with Np rows and D1 columns and a first other label with Np rows. The first acquired data includes a first feature quantity matrix with N1 rows and D1 columns and a first acquired label with N1 rows. The first generated data includes a first generated matrix with (Np+N1) rows and (3×D1) columns and a first generated label with (Np+N1) rows. (Np+N1) / D1 is equal to or larger than 250.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] FIELD Embodiments of the present invention relate to a data processing device, a data processing system, and a data processing method. [Background technology]

[0002] For example, data related to various electronic devices such as magnetic recording and reproducing devices is processed. For example, machine learning is performed by data processing. High-precision data processing is desired. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] Daume'III, H., "Frustratingly Easy Domain Adaptation," Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics, pp. 256-263, June 2007. Summary of the Invention [Problem to be solved by the invention]

[0004] The embodiments of the present invention provide a data processing device, a data processing system, and a data processing method that are capable of high-precision data processing. [Means for solving the problem]

[0005] According to an embodiment of the present invention, a data processing device includes an acquisition unit and a processing unit. The acquisition unit is capable of acquiring first acquired data in a first operation. The processing unit is capable of generating a first machine learning model based on first generated data that is based on the first acquired data and first other data in the first operation. The first other data includes a first other feature matrix of Np rows and D1 columns and a first other label of Np rows. Np is an integer equal to or greater than 2. D1 is an integer equal to or greater than 1. The first acquired data includes a first feature matrix of N1 rows and D1 columns and a first acquired label of N1 rows. N1 is an integer equal to or greater than 2. N1 is smaller than Np. The first generated data includes a first generator matrix of (Np+N1) rows and (3×D1) columns and a first generated label of (Np+N1) rows. The first generator matrix includes first matrix data, second matrix data, and third matrix data. The elements of the first matrix data include a row-wise combination of the first other feature matrix and the first feature matrix. The elements of the second matrix data include a row-wise combination of an Np-row, D1-column matrix of 0 elements and the first feature matrix. The elements of the third matrix data include a row-wise combination of the first other feature matrix and an N1-row, D1-column matrix of 0 elements. The elements of the first generated label include a row-wise combination of the first other label and the first obtained label. (Np+N1) / D1 is 250 or greater. [Brief explanation of the drawings]

[0006] [Figure 1] FIG. 1 is a schematic diagram illustrating a data processing device according to the first embodiment. [Figure 2] FIG. 2 is a schematic diagram illustrating the operation of the data processing device according to the first embodiment. [Figure 3] FIG. 3 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 4] FIG. 4 is a schematic diagram illustrating the operation of the data processing device according to the first embodiment. [Figure 5] FIG. 5 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 6] FIG. 6 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 7] FIG. 7 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 8] FIG. 8 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 9] FIG. 9 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 10] FIG. 10 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 11] FIG. 11 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 12] FIG. 12 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 13] FIG. 13 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 14] FIG. 14 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 15] FIG. 15 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 16] FIG. 16 is a schematic view illustrating the operation of the data processing device according to the first embodiment. [Figure 17] FIG. 17 is a graph illustrating the characteristics of the data processing device according to the first embodiment. [Figure 18] FIG. 18 is a schematic diagram illustrating a data processing device according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0007] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The drawings are schematic or conceptual. In the present specification and the drawings, elements similar to those described above with reference to the previous drawings are designated by the same reference numerals, and detailed descriptions thereof will be omitted where appropriate.

[0008] (First embodiment) FIG. 1 is a schematic diagram illustrating a data processing device according to the first embodiment. As shown in FIG. 1 , a data processing device 110 according to the embodiment includes a processing unit 71. The processing unit 71 is capable of acquiring various types of data 10D. For example, the data processing device 110 may include an acquisition unit 72. The acquisition unit 72 may acquire various types of data 10D, and the data 10D acquired by the acquisition unit 72 may be supplied to the processing unit 71. The data processing device 110 may include a storage unit 73. The data 10D acquired by the acquisition unit 72 may be stored in the storage unit 73. The processing unit 71 may acquire the data 10D stored in the storage unit 73 from the storage unit 73.

[0009] The acquisition unit 72 is, for example, an interface. The acquisition unit 72 may be, for example, an interface for input and output. The processing unit 71 may output information I1 related to the processing result. The information I1 may be output via the acquisition unit 72 (interface). The processing unit 71 may be capable of communicating with the server 74. The communication may include at least one of providing information and obtaining information. The communication may be based on any wired or wireless method.

[0010] The data 10D may include, for example, acquired data (for example, first acquired data 11 and second acquired data 12) etc. The data 10D may also include other data (for example, first other data 51 and second other data 52) etc.

[0011] FIG. 2 is a schematic diagram illustrating the operation of the data processing device according to the first embodiment. In the embodiment, the acquisition unit 72 and the processing unit 71 are capable of performing a first operation OP1. The acquisition unit 72 is capable of acquiring first acquired data 11 in the first operation OP1. The processing unit 71 may acquire the first acquired data 11 from the acquisition unit 72. The processing unit 71 may acquire the first acquired data 11 stored in the memory unit 73.

[0012] The processing unit 71 acquires the first other data 51. For example, the acquisition unit 72 may acquire the first other data 51, and the processing unit 71 may acquire the first other data 51 from the acquisition unit 72. The processing unit 71 may acquire the first other data 51 stored in the storage unit 73.

[0013] In the first operation OP1, the processing unit 71 can generate the first machine learning model 31 based on the first generated data 21 that is based on the first acquired data 11 and the first other data 51.

[0014] The first other data 51 includes a first other feature matrix 51a with Np rows and D1 columns, and a first other label 51b with Np rows. "Np" is an integer equal to or greater than 2. The first other label 51b corresponds to the first other feature matrix 51a. "Np" corresponds to, for example, the number of samples in the first other data 51.

[0015] The first acquired data 11 includes a first feature matrix 11a with N1 rows and D1 columns and a first acquired label 11b with N1 rows. "N1" is an integer equal to or greater than 2. "D1" is an integer equal to or greater than 1. "N1" is smaller than "Np". "N1" corresponds to, for example, the number of samples in the first acquired data 11.

[0016] In one example, the first acquired data 11 is small-scale data related to the target device. The first other data 51 is large-scale data related to at least one of the target device and devices similar to the target device. For example, the first acquired data 11 relates to evaluation data related to a small-scale experiment. For example, the first other data 51 relates to evaluation data related to a mass-produced product.

[0017] In one example, the target device is a magnetic recording and reproducing device. The first acquired data 11 relates to data on a prototype of the magnetic recording and reproducing device. The first other data 51 relates to data on a mass-produced version of the magnetic recording and reproducing device.

[0018] The processing unit 71 can generate first generated data 21 based on such first acquired data 11 and first other data 51. As shown in FIG. 2, the first generated data 21 includes a first generated matrix 21a and a first generated label 21b. The first generated matrix 21a is a matrix with (Np+N1) rows and (3×D1) columns. The first generated label 21b is data with (Np+N1) rows. The number of columns of the first generated label 21b is, for example, 1.

[0019] The first generator matrix 21a includes first matrix data M1, second matrix data M2, and third matrix data M3.

[0020] The elements of the first matrix data M1 include a combination in the row direction of the first other feature amount matrix 51a and the first feature amount matrix 11a.

[0021] The elements of the second matrix data M2 include a row-wise combination of a matrix Mxa1 of elements of 0 with Np rows and D1 columns (ie, a zero matrix) and the first feature amount matrix 11a.

[0022] The elements of the third matrix data M3 include a row-wise combination of the first other feature amount matrix 51a and a matrix Mxb1 of 0 elements with N1 rows and D1 columns (ie, a 0 matrix).

[0023] The components of the first generated label 21b include a combination in the row direction between the first other label 51b and the first obtained label 11b.

[0024] In an embodiment, (Np+N1) / D1 is, for example, equal to or greater than 250. (Np+N1) / D1 is defined as a first ratio R1. As will be described later, in an embodiment, the first ratio R1 may be equal to or greater than 500.

[0025] In the embodiment, such first generated data 21 is generated. A first machine learning model 31 is generated based on the first generated data 21. In the first machine learning model 31, highly accurate data processing becomes possible.

[0026] For example, in the embodiment, transfer learning is performed. In the transfer learning according to the embodiment, for example, first generated data 21 is derived by combining first acquired data 11 (target data) with first other data 51. A machine learning model based on such first generated data 21 is used. This achieves higher accuracy than the first reference example in which a machine learning model based only on the target data is used.

[0027] As will be described later, higher accuracy than that of the first reference example can be obtained when the first ratio R1 (that is, (Np+N1) / D1) is equal to or greater than 250. An example of the relationship between the first ratio R1 and accuracy will be described later.

[0028] 3 to 5 are schematic views illustrating the operation of the data processing device according to the first embodiment. Fig. 3 illustrates the first other data 51. Fig. 4 illustrates the first acquired data 11. Fig. 5 illustrates the first generated data 21.

[0029] 3, the first other data 51 includes Np rows of first other labels 51b. The first other labels 51b include, for example, yp_1; yp_2; ... yp_Np. The first other data 51 includes Np rows and D1 columns of a first other feature matrix 51a. The first row of the first other feature matrix 51a includes, for example, xp_1,1; xp_1,2; ... xp_1,D1. The second row of the first other feature matrix 51a includes, for example, xp_2,1; xp_2,2; ... xp_2,D1. The Np-th row of the first other feature matrix 51a includes, for example, xp_Np,1; xp_Np,2; ... xp_Np,D1.

[0030] 4, the first acquired data 11 includes first acquired labels 11b with N1 rows. The first acquired labels 11b include, for example, y1_1; y1_2; ... y1_N1. The first acquired data 11 includes a first feature matrix 11a with N1 rows and D1 columns. The first row of the first feature matrix 11a includes, for example, x1_1,1; x1_1,2; ... x1_1,D1. The second row of the first feature matrix 11a includes, for example, x1_2,1; x1_2,2; ... x1_2,D1. The N1 row of the first feature matrix 11a includes, for example, x1_N1,1; x1_N1,2; ... x1_N1,D1.

[0031] 5, the first generated data 21 includes first generated labels 21b of (Np+N1) rows. The first generated labels 21b include, for example, yp_1; yp_2; ... yp_Np; y1_1; y1_2; ... y1_N1.

[0032] The first generated data 21 includes a first generator matrix 21a with (Np+N1) rows and (3×D1) columns.

[0033] The first row of the first generator matrix 21a includes, for example, xp_1,1; xp_1,2; ... xp_1,D1; D1 zeros; xp_1,1; xp_1,2; ... xp_1,D1. The second row of the first generator matrix 21a includes, for example, xp_2,1; xp_2,2; ... xp_2,D1; D1 zeros; xp_2,1; xp_2,2; ... xp_2,D1. The Np-th row of the first generator matrix 21a includes, for example, xp_Np,1; xp_Np,2; ... xp_Np,D1; D1 zeros; xp_Np,1; xp_Np,2; ... xp_Np,D1.

[0034] The (Np+1)-th row of the first generator matrix 21a includes, for example, x1_1,1; x1_1,2; ... x1_1,D1; x1_1,1; x1_1,2; ... x1_1,D1;D1 0s. The (Np+2)-th row of the first generator matrix 21a includes, for example, x1_2,1; x1_2,2; ... x1_2,D1; x1_2,1; x1_2,2; ... x1_2,D1;D1 0s. The (Np+N1)-th row of the first generator matrix 21a includes, for example, x1_N1,1; x1_N1,2; ... x1_N1,D1; x1_N1,1; x1_N1,2; ... x1_N1,D1;D1 0s.

[0035] A first machine learning model 31 is generated based on such first generated data 21. For example, the processing unit 71 generates the first machine learning model 31 by at least one selected from the group consisting of kernel regression, linear regression, ridge regression, lasso regression, elastic net, gradient boosting regression, random forest regression, k-nearest neighbor regression, and logistic regression.

[0036] Kernel regression may include, for example, Gaussian process regression and / or Support Vector Regression (SVR).

[0037] 1, in the embodiment, the storage unit 73 may include a first storage area 73a and a second storage area 73b. For example, the first acquired data 11 acquired by the acquisition unit 72 is stored in the first storage area 73a. For example, the first other data 51 may be stored in the first storage area 73a. The processing unit 71 may acquire the first acquired data 11 and the first other data 51 from the first storage area 73a and perform the first operation OP1.

[0038] 1, the storage unit 73 may include a second storage area 73b. The processing unit 71 may be capable of storing the generated first generated data 21 in the second storage area 73b. The processing unit 71 may be capable of storing the derived first machine learning model 31 in the second storage area 73b.

[0039] The processing unit 71 may be further capable of performing the following processes in the first operation OP1. FIG. 6 is a schematic view illustrating the operation of the data processing device according to the first embodiment. As shown in FIG. 6, in a first operation OP1, the processing unit 71 can input the first regression matrix 61a to the first machine learning model 31 to further derive a first regression label 61b.

[0040] The first regression matrix 61a has N1 rows and (3×D1) columns. The first regression matrix 61a includes first regression matrix data K1, second regression matrix data K2, and third regression matrix data K3. The components of the first regression matrix data K1 include the first feature amount matrix 11a. The components of the second regression matrix data K2 include the first feature amount matrix 11a. The components of the third regression matrix data K3 include a matrix Mxc1 of zero elements with N1 rows and D1 columns (i.e., a zero matrix). The derived first regression label 61b has N1 rows.

[0041] The first regression label 61b obtained in this manner has high accuracy. For example, in the first reference example described above, a machine learning model based on the first acquired data 11 (target data) is used. In the first reference example, the accuracy of the regression label obtained using this machine learning model is low. In the embodiment, the first regression label 61b obtained has higher accuracy than that of the first reference example.

[0042] FIG. 7 is a schematic view illustrating the operation of the data processing device according to the first embodiment. 7 illustrates a first regression matrix 61a. The first row of the first regression matrix 61a includes, for example, x1_1,1; x1_1,2; ... x1_1,D1; x1_1,1; x1_1,2; ... x1_1,D1; D1 zeros. The second row of the first regression matrix 61a includes, for example, x1_2,1; x1_2,2; ... x1_2,D1; x1_2,1; x1_2,2; ... x1_2,D1; D1 zeros. The N1 row of the first regression matrix 61a includes, for example, x1_N1,1; x1_N1,2; ... x1_N1,D1; x1_N1,1; x1_N1,2; ... x1_N1,D1; D1 zeros. Such a first regression matrix 61a is input to the first machine learning model 31 to obtain first regression labels 61b.

[0043] FIG. 8 is a schematic view illustrating the operation of the data processing device according to the first embodiment. Fig. 8 illustrates the first regression label 61b. As shown in Fig. 8, the first regression label 61b includes, for example, z1_1; z1_2; ... z1_N1. Such a first regression label 61b has high accuracy.

[0044] As will be explained below, the acquisition unit 72 and the processing unit 71 may be capable of performing a second operation.

[0045] FIG. 9 is a schematic view illustrating the operation of the data processing device according to the first embodiment. In the embodiment, the acquisition unit 72 and the processing unit 71 are capable of performing a second operation OP2. The acquisition unit 72 is capable of acquiring second acquired data 12 in the second operation OP2. The processing unit 71 may acquire the second acquired data 12 from the acquisition unit 72. The processing unit 71 may acquire the second acquired data 12 stored in the memory unit 73.

[0046] The processing unit 71 acquires the second other data 52. For example, the acquisition unit 72 may acquire the second other data 52, and the processing unit 71 may acquire the second other data 52 from the acquisition unit 72. The processing unit 71 may acquire the second other data 52 stored in the storage unit 73.

[0047] In the second operation OP2, the processing unit 71 can generate the second machine learning model 32 based on the second generated data 22 that is based on the second acquired data 12 and the second other data 52.

[0048] The second other data 52 includes a second other feature matrix 52a with Nq rows and D2 columns, and a second other label 52b with Nq rows. "Nq" is an integer equal to or greater than 2. The second other label 52b corresponds to the second other feature matrix 52a. "Nq" corresponds to, for example, the number of samples in the second other data 52.

[0049] The second acquired data 12 includes a second feature matrix 12a with N2 rows and D2 columns and a second acquired label 12b with N2 rows. "N2" is an integer equal to or greater than 2. "D2" is an integer equal to or greater than 1. "N2" is smaller than "Nq." "N2" corresponds to, for example, the number of samples in the second acquired data 12.

[0050] For example, the second obtained data 12 is small-scale data related to the target device corresponding to the second obtained data 12. The second other data 52 is large-scale data related to at least one of the target device and devices similar to the target device.

[0051] The processing unit 71 can generate second generated data 22 based on such second acquired data 12 and second other data 52. As shown in Fig. 9, the second generated data 22 includes a second generated matrix 22a and a second generated label 22b. The second generated matrix 22a is a matrix with (Nq+N2) rows and (3 x D2) columns. The second generated label 22b is data with (Nq+N2) rows. The number of columns of the second generated label 22b is, for example, 1.

[0052] The second generator matrix 22a includes fourth matrix data M4, fifth matrix data M5, and sixth matrix data M6.

[0053] The elements of the fourth matrix data M4 include a combination in the row direction of the second other feature amount matrix 52a and the second feature amount matrix 12a.

[0054] The elements of the fifth matrix data M5 include a row-wise combination of a matrix Mxa2 of elements of 0 with Nq rows and D2 columns (ie, a zero matrix) and the second feature amount matrix 12a.

[0055] The elements of the sixth matrix data M6 include a row-wise combination of the second other feature amount matrix 52a and a matrix Mxb2 of 0 elements with N2 rows and D2 columns (ie, a 0 matrix).

[0056] The components of the second generated label 22b include a combination in the row direction between the second other label 52b and the second obtained label 12b.

[0057] In the second operation OP2, (Nq+N2) / D2 is, for example, equal to or greater than 250. This ratio may be equal to or greater than 500.

[0058] In the embodiment, such second generated data 22 is generated. A second machine learning model 32 is generated based on the second generated data 22. In the second machine learning model 32, highly accurate data processing becomes possible.

[0059] 10 to 12 are schematic views illustrating the operation of the data processing device according to the first embodiment. Fig. 10 illustrates the second other data 52. Fig. 11 illustrates the second acquired data 12. Fig. 12 illustrates the second generated data 22.

[0060] 10, the second other data 52 includes Nq rows of second other labels 52b. The second other labels 52b include, for example, yq_1; yq_2; ... yq_Np. The second other data 52 includes a second other feature matrix 52a with Nq rows and D2 columns. The first row of the second other feature matrix 52a includes, for example, xq_1,1; xq_1,2; ... xq_1,D2. The second row of the second other feature matrix 52a includes, for example, xq_2,1; xq_2,2; ... xq_2,D2. The Nq-th row of the second other feature matrix 52a includes, for example, xq_Nq,1; xq_Nq,2; ... xq_Nq,D2.

[0061] 11, the second acquired data 12 includes second acquired labels 12b with N2 rows. The second acquired labels 12b include, for example, y2_1; y2_2; ... y2_N2. The second acquired data 12 includes a second feature matrix 12a with N2 rows and D2 columns. The first row of the second feature matrix 12a includes, for example, x2_1,1; x2_1,2; ... x2_1,D2. The second row of the second feature matrix 12a includes, for example, x2_2,1; x2_2,2; ... x2_2,D2. The N2 row of the second feature matrix 12a includes, for example, x2_N2,1; x2_N2,2; ... x2_N2,D2.

[0062] 12, the second generated data 22 includes second generated labels 22b of (Nq+N2) rows. The second generated labels 22b include, for example, yq_1; yq_2; ... yq_Nq; y2_1; y2_2; ... y2_N2.

[0063] The second generator data 22 includes a second generator matrix 22a with (Nq+N2) rows and (3×D2) columns.

[0064] The first row of the second generator matrix 22a includes, for example, xq_1,1; xq_1,2; ... xq_1,D2; D2 zeros; xq_1,1; xq_1,2; ... xq_1,D2. The second row of the second generator matrix 22a includes, for example, xp_2,1; xp_2,2; ... xp_2,D2; D2 zeros; xq_2,1; xq_2,2; ... xq_2,D2. The Nq-th row of the second generator matrix 22a includes, for example, xq_Np,1; xq_Np,2; ... xq_Np,D2; D2 zeros; xq_Nq,1; xq_Np,2; ... xq_Nq,D2.

[0065] The (Nq+1)-th row of the second generator matrix 22a includes, for example, x2_1,1; x2_1,2; ... x2_1,D2; x2_1,1; x2_1,2; ... x2_1,D2; D2 0s. The (Nq+2)-th row of the second generator matrix 22a includes, for example, x2_2,1; x2_2,2; ... x2_2,D2; x2_2,1; x2_2,2; ... x2_2,D2; D2 0s. The (Nq+N2)-th row of the second generator matrix 22a includes, for example, x2_N2,1; x2_N2,2; ... x2_N2,D2; x2_N2,1; x2_N2,2; ... x2_N2,D2; D2 0s.

[0066] A second machine learning model 32 is generated based on such second generation data 22. The second machine learning model 32 may be generated in the same manner as the first machine learning model 31, for example.

[0067] The second acquired data 12 may be stored in the first storage area 73 a. For example, the second other data 52 may be stored in the first storage area 73 a. The processing unit 71 may acquire the second acquired data 12 and the second other data 52 from the first storage area 73 a and perform the second operation OP2.

[0068] The processing unit 71 may be capable of storing the generated second generated data 22 in the second storage area 73 b. The processing unit 71 may be capable of storing the derived second machine learning model 32 in the second storage area 73 b.

[0069] The processing unit 71 may further be able to perform the following processes in the second operation OP2. FIG. 13 is a schematic view illustrating the operation of the data processing device according to the first embodiment. As shown in FIG. 13, in a second operation OP2, the processing unit 71 can input the second regression matrix 62a to the second machine learning model 32 to further derive a second regression label 62b.

[0070] The second regression matrix 62a has N2 rows and (3×D2) columns. The second regression matrix 62a includes fourth regression matrix data K4, fifth regression matrix data K5, and sixth regression matrix data K6. The components of the fourth regression matrix data K4 include the second feature amount matrix 12a. The components of the fifth regression matrix data K5 include the second feature amount matrix 12a. The components of the sixth regression matrix data K6 include a matrix Mxc2 of zero elements with N2 rows and D2 columns (i.e., a zero matrix). The derived second regression label 62b has N2 rows.

[0071] The second regression label 62b obtained in this manner has high accuracy. For example, the accuracy of the second regression label 62b is high. For example, in the first reference example, a machine learning model based on the second acquired data 12 (target data) is used. In the first reference example, the accuracy of the regression label obtained using this machine learning model is low. In the embodiment, the second regression label 62b having higher accuracy than that of the first reference example is obtained.

[0072] FIG. 14 is a schematic view illustrating the operation of the data processing device according to the first embodiment. 14 illustrates an example of the second regression matrix 62a. The first row of the second regression matrix 62a includes, for example, x2_1,1; x2_1,2; ... x2_1,D2; x2_1,1; x2_1,2; ... x2_1,D2; D2 zeros. The second row of the second regression matrix 62a includes, for example, x2_2,1; x2_2,2; ... x2_2,D2; x2_2,1; x2_2,2; ... x2_2,D2; D2 zeros. The N2-th row of the second regression matrix 62a includes, for example, x2_N1,1; x2_N1,2; ... x2_N2,D2; x2_N2,1; x2_N2,2; ... x2_N2,D2; D2 zeros. Such a second regression matrix 62a is input to the second machine learning model 32 to obtain second regression labels 62b.

[0073] FIG. 15 is a schematic view illustrating the operation of the data processing device according to the first embodiment. Fig. 15 illustrates an example of the second regression label 62b. As shown in Fig. 15, the second regression label 62b includes, for example, z2_1; z2_2; ... z2_N2. Such second regression label 62b has high accuracy.

[0074] The processing unit 71 may be capable of performing a third operation. FIG. 16 is a schematic view illustrating the operation of the data processing device according to the first embodiment. As shown in FIG. 16, in the third operation OP3, the processing unit 71 outputs information I1 relating to the comparison between the first regression label 61b and the second regression label 62b.

[0075] For example, the information I1 includes at least one of a first result and a second result. The first result includes a comparison result between the maximum value of the component of the first regression label 61b and the maximum value of the component of the second regression label 62b. The second result includes a comparison result between the minimum value of the component of the first regression label 61b and the minimum value of the component of the second regression label 62b.

[0076] For example, if the target device is a magnetic recording / reproducing device, the feature value includes the head width of the magnetic head. The label includes, for example, areal recording density. For example, the first regression label 61b includes multiple values ​​related to areal recording density. The second regression label 62b includes multiple values ​​related to areal recording density. For example, the maximum value of the multiple values ​​included in the first regression label 61b is compared with the maximum value of the multiple values ​​included in the second regression label 62b. Information I1 regarding the result of the comparison is output.

[0077] For example, the first acquired data 11 is data related to a first prototype. The second acquired data 12 is data related to a second prototype. The conditions for the second prototype are different from the conditions for the first prototype. In the embodiment, the results of multiple prototypes can be evaluated with higher accuracy.

[0078] An example of the characteristics of the data processing device 110 according to the embodiment will be described below. The following example relates to the areal recording density of two types of magnetic recording and reproducing devices under different conditions. A sufficiently large amount of data has already been obtained to enable accurate judgment for each of the two types of magnetic recording and reproducing devices. That is, first population data has already been obtained for the magnetic recording and reproducing device under the first condition, and second population data has already been obtained for the magnetic recording and reproducing device under the second condition. From the first population data and the second population data, the relative merits ("actual ranking") of performance (in this example, areal recording density) under the first and second conditions are known. These data are used experimentally to evaluate the characteristics of the data processing device.

[0079] In order to evaluate the characteristics of the data processing device, arbitrary data is extracted from the first population data to obtain first acquired data 11. The number of first acquired data 11 is smaller than the number of first population data. On the other hand, arbitrary data is extracted from the second population data to obtain second acquired data 12. The number of second acquired data 12 is smaller than the number of second population data.

[0080] In the first calculation example described below, first generated data 21 is generated based on first other data 51 and first acquired data 11. Furthermore, second generated data 22 is generated based on second other data 52 and second acquired data 12. The first other feature matrix 51a, the first feature matrix 11a, the second other feature matrix 52a, and the second feature matrix 12a represent the head widths of the magnetic heads in the magnetic recording and reproducing device. The first other label 51b, the first acquired label 11b, the second other label 52b, and the second acquired label 12b represent the areal recording densities of the magnetic recording and reproducing device.

[0081] Furthermore, a first machine learning model 31 is derived. Furthermore, a first regression label 61b is derived by inputting the first regression matrix 61a into the first machine learning model 31. The first regression label 61b thus obtained may not be accurate because the number of first obtained data 11 is smaller than the number of first population data.

[0082] Meanwhile, a second machine learning model 32 is derived. Furthermore, a second regression label 62b is derived by inputting the second regression matrix 62a into the second machine learning model 32. The second regression label 62b thus obtained may not be accurate because the number of second obtained data 12 is smaller than the number of second population data.

[0083] The first regression label 61b and the second regression label 62b are evaluated for superiority or inferiority. If the evaluation result of the superiority or inferiority of the regression labels is the same as the superiority or inferiority based on the first population data and the second population data, it corresponds to a "correct answer." If the evaluation result of the superiority or inferiority of the regression labels is different from the superiority or inferiority based on the first population data and the second population data, it corresponds to an "incorrect answer."

[0084] In the evaluation described below, the first ratio R1 (i.e., (Np+N1) / D1) is changed to derive the "correct answer rate." In the following example, "D1" is 1. "Nq" is the same as "Np." "D2" is the same as "D1." (Nq+N2) / D2 is the same as (Np+N1) / D1.

[0085] When the first ratio R1 is higher than 500, "N1" and "N2" are kept constant at 200, and the first ratio R1 is changed by changing "Np." On the other hand, when the first ratio R1 is 500 or less, Np / N1 is kept constant at 1.5, and the first ratio R1 is changed by changing "Np" and "N1."

[0086] On the other hand, in the second calculation example described below, a machine learning model based only on the first acquired data 11 and a machine learning model based only on the second acquired data 12 are used. In the second calculation example, the first other data 51 is not used, and the second other data 52 is not used. In these calculation examples, the machine learning model is linear regression. In the second calculation example as well, the merits of the first regression label 61b and the second regression label 62b are evaluated, and the "correct answer rate" is calculated.

[0087] FIG. 17 is a graph illustrating the characteristics of the data processing device according to the first embodiment. 17 shows the results of the first calculation example CC1 and the second calculation example CC2. As already explained, in the first calculation example CC1, a first machine learning model 31 based on the first generated data 21 and a second machine learning model 32 based on the second generated data 22 are used by the above-described method according to the embodiment. On the other hand, in the second calculation example CC2, a machine learning model based only on the first acquired data 11 and a machine learning model based only on the second acquired data 12 are used. The second calculation example CC2 corresponds to the above-described first reference example.

[0088] The horizontal axis of FIG. 17 is the first ratio R1. As already explained, the first ratio R1 is (Np+N1) / D1. The vertical axis of FIG. 17 is the accuracy A1 of the derived first regression label 61b and second regression label 62b. The accuracy A1 corresponds to the accuracy rate of the ranking of the result obtained by calculation with respect to the actual ranking. The ranking corresponds to the superiority or inferiority of the areal recording density of the two types of magnetic recording and reproducing devices.

[0089] As shown in Figure 17, in the second calculation example CC2, the accuracy A1 is 0.805. In the first calculation example CC1, the accuracy A1 increases as the first ratio R1 increases. In the first calculation example CC1, when the first ratio R1 is 250 or greater, the accuracy A1 exceeds the accuracy A1 in the second calculation example CC2. In the first calculation example CC1, when the first ratio R1 is 500 or greater, a higher accuracy A1 of 0.9 or greater is obtained.

[0090] In the embodiment, the first ratio R1 is preferably 250 or more, which allows a high accuracy A1 to be obtained. The first ratio R1 is more preferably 500 or more, which allows a higher accuracy A1 to be obtained.

[0091] For example, when comparing feature quantities under different conditions, as in the above-described third operation OP3, if the accuracy A1 of the machine learning model is low, it is difficult to make a correct comparison. In the embodiment, by setting the first ratio R1 to 250 or more (or, for example, 500 or more), feature quantities under different conditions can be compared with high accuracy A1. For example, the first generated label 21b and the second generated label 22b can be compared and evaluated with high accuracy.

[0092] As described above, in the embodiment, a machine learning model is used that is based on the first generated data 21 in which the first acquired data 11 (target data) is combined with the first other data 51. This provides higher accuracy than the first reference example in which a machine learning model based only on the target data is used.

[0093] For example, if the target device is a magnetic recording / reproducing device, the characteristic amount is the head width of the magnetic head. In this case, the label may be, for example, the areal recording density.

[0094] For example, when developing a target device, if the number of samples is small, the regression accuracy may be low. In the embodiment, even if the number of samples is small, the characteristics of the developed item can be evaluated with high accuracy.

[0095] In this embodiment, it is preferable that Np / N1 is 1.5 or more. High accuracy can be obtained even with a small "N1".

[0096] In the embodiment, the first acquired data 11, the first other data 51, the second acquired data 12, and the second other data 52 may include characteristics of the magnetic recording and reproducing device.

[0097] The characteristics of the magnetic recording and reproducing device include, for example, at least one selected from the group consisting of SNR (Signal-Noise Ratio), BER (Bit Error Rate), Fringe BER, EWAC (Erase Width at AC erase), MWW (Magnetic Write track Width), OW (Over Write), SOVA-BER (Soft Viterbi Algorithm-BER), VMM (Viterbi Metric Margi), PRO (Repeatable RunOut), and NRRO (Non-Repeatable RunOut).

[0098] In the embodiment, "D1" may be 1. For example, a high first ratio R1 is obtained. With a smaller "D1", high-precision processing can be performed.

[0099] In the data processing system 210 (see FIG. 1) according to the embodiment, multiple elements (e.g., the processing unit 71 and the storage unit 73) may be provided in different locations. Information may be transmitted and received by any communication method. For example, multiple parts included in the processing unit 71 may be provided in different locations.

[0100] For example, the data processing system 210 may include one or more acquisition units 72 and one or more processing units 71. A portion of the first operation OP1 may be performed by a portion of one or more processing units 71. Another portion of the first operation OP1 may be performed by another portion of the one or more processing units 71. At least a portion of the second operation OP2 and the third operation OP3 may be performed by another portion of the one or more processing units 71.

[0101] FIG. 18 is a schematic diagram illustrating a data processing device according to the embodiment. 18, the data processing device 110 includes a processing unit 71, an acquisition unit 72, and a storage unit 73. The processing unit 71 is, for example, an electric circuit. The storage unit 73 may include, for example, at least one of a read-only memory (ROM) and a random access memory (RAM). Any storage device may be used as the storage unit 73.

[0102] The data processing device 110 may include a display unit 79b and an input unit 79c. The display unit 79b may include various types of displays. The input unit 79c includes, for example, a device with an operation function (such as a keyboard, a mouse, a touch-type input panel, or a voice recognition input device).

[0103] The embodiment may include a program that causes a computer (processing unit 71) to perform the above-described operations. The embodiment may include a storage medium on which the program is stored.

[0104] (Second embodiment) The second embodiment relates to a data processing method, which causes a processing unit 71 to perform a first operation OP1. In the first operation OP1, the processing unit 71 can generate a first machine learning model 31 based on first generated data 21 that is based on first acquired data 11 and first other data 51.

[0105] The first other data 51 includes a first other feature matrix 51a with Np rows and D1 columns, and a first other label 51b with Np rows. "Np" is an integer greater than or equal to 2. The first acquired data 11 includes a first feature matrix 11a with N1 rows and D1 columns, and a first acquired label 11b with N1 rows. "N1" is an integer greater than or equal to 2. "D1" is an integer greater than or equal to 1. "N1" is smaller than "Np".

[0106] The first generated data 21 includes a first generated matrix 21a with (Np+N1) rows and (3×D1) columns, and a first generated label 21b with (Np+N1) rows. The first generated matrix 21a includes first matrix data M1, second matrix data M2, and third matrix data M3.

[0107] The elements of the first matrix data M1 include row-wise combinations of the first other feature matrix 51a and the first feature matrix 11a. The elements of the second matrix data M2 include row-wise combinations of a matrix Mxa1 of elements with zeros in Np rows and D1 columns and the first feature matrix 11a. The elements of the third matrix data M3 include row-wise combinations of the first other feature matrix 51a and a matrix Mxb1 of elements with zeros in N1 rows and D1 columns.

[0108] The component of the first generated label 21b includes a combination in the row direction between the first other label 51b and the first obtained label 11b. (Np+N1) / D1 is 250 or more. This ratio may be 500 or more.

[0109] The embodiment may include the following configurations (e.g., technical solutions). (Configuration 1) an acquisition unit; a processing unit, the acquisition unit is capable of acquiring first acquisition data in a first operation; In the first operation, the processing unit is capable of generating a first machine learning model based on first generated data that is based on the first acquired data and first other data; the first other data includes a first other feature matrix of Np rows and D1 columns and a first other label of Np rows, where Np is an integer equal to or greater than 2 and D1 is an integer equal to or greater than 1; the first acquired data includes a first feature matrix of N1 rows and D1 columns and a first acquired label of N1 rows, where N1 is an integer equal to or greater than 2 and is smaller than Np; the first generated data includes a first generator matrix of (Np+N1) rows and (3×D1) columns and a first generated label of (Np+N1) rows; the first generator matrix includes first matrix data, second matrix data, and third matrix data; an element of the first matrix data includes a combination of the first other feature amount matrix and the first feature amount matrix in a row direction; the elements of the second matrix data include a combination in the row direction of a matrix of elements of 0 in Np rows and D1 columns and the first feature amount matrix, the elements of the third matrix data include a combination in the row direction of the first other feature amount matrix and a matrix of elements of 0 with N1 rows and D1 columns, a component of the first generated label includes a combination of the first other label and the first obtained label in the row direction; A data processing device, wherein (Np+N1) / D1 is 250 or greater.

[0110] (Configuration 2) Further comprising a storage unit, the storage unit includes a first storage area, the first acquired data and the first other data are stored in the first other storage area; 2. The data processing device according to configuration 1, wherein the processing unit is capable of acquiring the first acquired data and the first other data from the first storage area and performing the first operation.

[0111] (Configuration 3) the storage unit further includes a second storage area; 3. The data processing device according to configuration 2, wherein the processing unit is capable of storing the first generated data in the second storage area.

[0112] (Configuration 4) In the first operation, the processing unit is capable of inputting a first regression matrix into the first machine learning model to further derive a first regression label; The first regression matrix has N1 rows and (3×D1) columns; the first regression matrix includes first regression matrix data, second regression matrix data, and third regression matrix data; components of the first regression matrix data include the first feature matrix; components of the second regression matrix data include the first feature matrix; 4. The data processing device according to any one of configurations 1 to 3, wherein the elements of the third regression matrix data include a matrix of 0 elements with N1 rows and D1 columns.

[0113] (Configuration 5) 5. The data processing device of configuration 4, wherein the first regression label is N1 rows.

[0114] (Configuration 6) the acquisition unit and the processing unit are further capable of performing a second operation; the acquisition unit is capable of acquiring second acquisition data in a second operation; In the second operation, the processing unit is capable of generating a second machine learning model based on second generated data that is based on the second acquired data and second other data; the second other data includes a second other feature matrix of Nq rows and D2 columns and a second other label of Np rows, where Np is an integer equal to or greater than 2 and D2 is an integer equal to or greater than 1; the second acquired data includes a second feature matrix of N2 rows and D2 columns and a second acquired label of N2 rows, where N2 is an integer equal to or greater than 2 and is smaller than Nq; the second generated data includes a second generator matrix with (Nq+N2) rows and (D2×3) columns and a second generated label with (Nq+N2) rows; the second generator matrix includes fourth matrix data, fifth matrix data, and sixth matrix data; an element of the fourth matrix data includes a combination of the second other feature amount matrix and the second feature amount matrix in a row direction; the elements of the fifth matrix data include a combination in the row direction of a matrix of elements of 0 in Nq rows and D2 columns and the second feature amount matrix, the elements of the sixth matrix data include a combination in the row direction of the second other feature amount matrix and a matrix of elements of 0 with N2 rows and D2 columns, a component of the second generated label includes a combination of the second other label and the second obtained label in the row direction; (Nq+N2) / D2 is 250 or more, In the second operation, the processing unit is capable of inputting a second regression matrix into the second machine learning model to further derive a second regression label; The second regression matrix has N rows (D × 3) columns, the second regression matrix includes fourth regression matrix data, fifth regression matrix data, and sixth regression matrix data; components of the fourth regression matrix data include the second feature matrix; components of the fifth regression matrix data include the second feature matrix; 6. The data processing device according to configuration 4 or 5, wherein the sixth regression matrix data elements include a matrix of zero elements with N2 rows and D2 columns.

[0115] (Configuration 7) The processing unit is further capable of performing a third operation; 7. The data processing device according to configuration 6, wherein the processing unit is capable of outputting information relating to the comparison between the first regression label and the second regression label in the third operation.

[0116] (Configuration 8) the information includes at least one of a first result and a second result; the first result includes a comparison result between a maximum value of a component of the first regression label and a maximum value of a component of the second regression label; 8. The data processing apparatus of claim 7, wherein the second result includes a comparison result between the minimum value of the component of the first regression label and the minimum value of the component of the second regression label.

[0117] (Configuration 9) 9. The data processing device according to any one of configurations 6 to 8, wherein the second regression label is N2 rows.

[0118] (Configuration 10) 10. The data processing device according to any one of configurations 1 to 9, wherein (Np+N1) / D1 is 500 or more.

[0119] (Configuration 11) 11. The data processing device according to any one of configurations 1 to 10, wherein Np / N1 is 1.5 or more.

[0120] (Configuration 12) 12. The data processing device according to any one of configurations 1 to 11, wherein the processing unit is capable of generating the first machine learning model by at least one selected from the group consisting of kernel regression, linear regression, ridge regression, lasso regression, elastic net, gradient boosting regression, random forest regression, k-nearest neighbor regression, and logistic regression.

[0121] (Configuration 13) 13. The data processing device according to configuration 12, wherein the kernel regression includes at least one of Gaussian process regression and SVR (Support Vector Regression).

[0122] (Configuration 14) 14. The data processing device according to any one of configurations 1 to 13, wherein the first obtained data, the first other data, and the second obtained data include characteristics of a magnetic recording and reproducing device.

[0123] (Configuration 15) 15. The data processing device of claim 14, wherein the characteristics include at least one selected from the group consisting of SNR (Signal-Noise Ratio), BER (Bit Error Rate), Fringe BER, EWAC (Erase Width at AC erase), MWW (Magnetic Write track Width), OW (Over Write), SOVA-BER (Soft Viterbi Algorithm-BER), VMM (Viterbi Metric Margi), PRO (Repeatable RunOut), and NRRO (Non-Repeatable RunOut).

[0124] (Configuration 16) 16. The data processing device according to any one of configurations 1 to 15, wherein D1 is 1.

[0125] (Configuration 17) one or more acquisition units; one or more processing units; the one or more acquisition units are capable of acquiring first acquisition data in a first operation; the one or more processing units are capable of generating a first machine learning model based on first generated data based on the first acquired data and first other data; the first other data includes a first other feature matrix of Np rows and D1 columns and a first other label of Np rows, where Np is an integer equal to or greater than 2 and D1 is an integer equal to or greater than 1; the first acquired data includes a first feature matrix of N1 rows and D1 columns and a first acquired label of N1 rows, where N1 is an integer equal to or greater than 2 and is smaller than Np; the first generated data includes a first generator matrix of (Np+N1) rows and (3×D1) columns and a first generated label of (Np+N1) rows; the first generator matrix includes first matrix data, second matrix data, and third matrix data; an element of the first matrix data includes a combination of the first other feature amount matrix and the first feature amount matrix in a row direction; the elements of the second matrix data include a combination in the row direction of a matrix of elements of 0 in Np rows and D1 columns and the first feature amount matrix, the elements of the third matrix data include a combination in the row direction of the first other feature amount matrix and a matrix of elements of 0 with N1 rows and D1 columns, a component of the first generated label includes a combination of the first other label and the first obtained label in the row direction; A data processing system, wherein (Np+N1) / D1 is 250 or greater.

[0126] (Configuration 18) a portion of the first operation is performed by a portion of the one or more processing units; 18. The data processing system of configuration 17, wherein another portion of the first operation is performed by another portion of the one or more processing units.

[0127] (Configuration 19) causing the processing unit to perform a first operation; In the first operation, the processing unit is capable of generating a first machine learning model based on first generated data that is based on the first acquired data and first other data; the first other data includes a first other feature matrix of Np rows and D1 columns and a first other label of Np rows, where Np is an integer equal to or greater than 2 and D1 is an integer equal to or greater than 1; the first acquired data includes a first feature matrix of N1 rows and D1 columns and a first acquired label of N1 rows, where N1 is an integer equal to or greater than 2 and is smaller than Np; the first generated data includes a first generator matrix of (Np+N1) rows and (3×D1) columns and a first generated label of (Np+N1) rows; the first generator matrix includes first matrix data, second matrix data, and third matrix data; an element of the first matrix data includes a combination of the first other feature amount matrix and the first feature amount matrix in a row direction; the elements of the second matrix data include a combination in the row direction of a matrix of elements of 0 in Np rows and D1 columns and the first feature amount matrix, the elements of the third matrix data include a combination in the row direction of the first other feature amount matrix and a matrix of elements of 0 with N1 rows and D1 columns, a component of the first generated label includes a combination of the first other label and the first obtained label in the row direction; A data processing method, wherein (Np+N1) / D1 is 250 or more.

[0128] According to the embodiments, it is possible to provide a data processing device, a data processing system, and a data processing method that are capable of performing highly accurate data processing.

[0129] The embodiments of the present invention have been described above with reference to examples. However, the present invention is not limited to these examples. For example, the specific configurations of the elements included in the data processing device, data processing system, and data processing method, such as the processing unit, acquisition unit, and storage unit, are within the scope of the present invention as long as a person skilled in the art can implement the present invention in a similar manner and obtain similar effects by appropriately selecting them from within the known range.

[0130] Any combination of two or more elements of each example within the scope of technical feasibility is also included within the scope of the present invention as long as it encompasses the gist of the present invention.

[0131] All data processing devices, data processing systems, and data processing methods that can be implemented by a person skilled in the art by appropriately modifying the design based on the data processing device, data processing system, and data processing method described above as embodiments of the present invention also fall within the scope of the present invention, as long as they include the gist of the present invention.

[0132] Within the scope of the concept of the present invention, a person skilled in the art may conceive of various modifications and alterations, and it is understood that these modifications and alterations also fall within the scope of the present invention.

[0133] Although several embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These novel embodiments can be embodied in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included within the scope and spirit of the invention, and are also included in the scope of the invention and its equivalents as defined in the claims. [Explanation of symbols]

[0134] 10D...data, 11, 12...first and second acquired data, 11a, 12a...first and second feature matrices, 11b, 12b...first and second acquired labels, 21, 22...first and second generated data, 21a, 22a...first and second generated matrices, 21b, 22b...first and second generated labels, 31, 32...first and second machine learning models, 51, 52...first and second other data, 51a, 52a...first and second other feature matrices, 51b, 52b...first and second other labels, 61a, 62a...first and second regression matrices, 61b, 62b...first and second regression labels, 71...processing unit, 72...acquisition unit, 73...storage unit, 73a, 73b...first and second storage areas, 74...server, 79b...display unit, 79c...input unit, 110...data processing device, 210...data processing system, A1...accuracy, CC1, CC2...first and second calculation examples, I1...information, K1 to K6...first to sixth regression matrix data, M1 to M6...first to sixth matrix data, Mxa1, Mxa2, Mxb1, Mxb2, Mxc1, Mxc2...matrices, OP1 to OP3...first to third operations, P1...information, R1...first ratio

Claims

1. an acquisition unit; a processing unit, the acquisition unit is capable of acquiring first acquisition data related to the target device in a first operation; In the first operation, the processing unit is capable of generating a first machine learning model based on first generated data that is based on the first acquired data and first other data; the first other data includes data related to a device similar to the target device, the first other data includes a first other feature matrix of Np rows and D1 columns and a first other label of Np rows, where Np is an integer equal to or greater than 2 and D1 is an integer equal to or greater than 1; the first acquired data includes a first feature matrix of N1 rows and D1 columns and a first acquired label of N1 rows, where N1 is an integer equal to or greater than 2 and is smaller than Np; the first generated data includes a first generator matrix of (Np+N1) rows and (3×D1) columns and a first generated label of (Np+N1) rows; the first generator matrix includes first matrix data, second matrix data, and third matrix data; an element of the first matrix data includes a combination of the first other feature amount matrix and the first feature amount matrix in a row direction; the elements of the second matrix data include a combination in the row direction of a matrix of elements of 0 in Np rows and D1 columns and the first feature amount matrix, the elements of the third matrix data include a combination in the row direction of the first other feature amount matrix and a matrix of elements of 0 with N1 rows and D1 columns, a component of the first generated label includes a combination of the first other label and the first obtained label in the row direction; A data processing device, wherein (Np+N1) / D1 is 250 or greater.

2. Further comprising a storage unit, the storage unit includes a first storage area, the first acquired data and the first other data are stored in the first storage area; The data processing device according to claim 1 , wherein the processing unit is capable of acquiring the first acquired data and the first other data from the first storage area and performing the first operation.

3. the storage unit further includes a second storage area; The data processing device according to claim 2 , wherein the processing unit is capable of storing the first generated data in the second storage area.

4. In the first operation, the processing unit is capable of inputting a first regression matrix into the first machine learning model to further derive a first regression label; The first regression matrix has N1 rows and (3×D1) columns; the first regression matrix includes first regression matrix data, second regression matrix data, and third regression matrix data; components of the first regression matrix data include the first feature amount matrix; components of the second regression matrix data include the first feature amount matrix; 2. The data processing apparatus according to claim 1, wherein the elements of said third regression matrix data include a matrix of zero elements with N1 rows and D1 columns.

5. the acquisition unit and the processing unit are further capable of performing a second operation; the acquisition unit is capable of acquiring second acquisition data in the second operation; In the second operation, the processing unit is capable of generating a second machine learning model based on second generated data that is based on the second acquired data and second other data; the second other data includes a second other feature matrix of Nq rows and D2 columns and a second other label of Nq rows, where Nq is an integer equal to or greater than 2 and D2 is an integer equal to or greater than 1; the second obtained data includes a second feature matrix of N2 rows and D2 columns and a second obtained label of N2 rows, where N2 is an integer equal to or greater than 2 and is smaller than Nq; the second generated data includes a second generator matrix with (Nq+N2) rows and (D2×3) columns and a second generated label with (Nq+N2) rows; the second generator matrix includes fourth matrix data, fifth matrix data, and sixth matrix data, an element of the fourth matrix data includes a combination of the second other feature amount matrix and the second feature amount matrix in a row direction; the elements of the fifth matrix data include a combination in the row direction of a matrix of elements of 0 in Nq rows and D2 columns and the second feature amount matrix, the elements of the sixth matrix data include a combination in the row direction of the second other feature amount matrix and a matrix of elements of 0 in N2 rows and D2 columns, a component of the second generated label includes a combination of the second other label and the second obtained label in the row direction; (Nq+N2) / D2 is 250 or more, In the second operation, the processing unit is capable of inputting a second regression matrix into the second machine learning model to further derive a second regression label; The second regression matrix has N rows (D × 3) columns, the second regression matrix includes fourth regression matrix data, fifth regression matrix data, and sixth regression matrix data; components of the fourth regression matrix data include the second feature amount matrix; components of the fifth regression matrix data include the second feature amount matrix; 5. The data processing apparatus according to claim 4, wherein the sixth regression matrix data elements include a matrix of zero elements with N2 rows and D2 columns.

6. The processing unit is further capable of performing a third operation; The data processing apparatus according to claim 5 , wherein the processing unit is capable of outputting information relating to the comparison between the first regression label and the second regression label in the third operation.

7. 7. The data processing device according to claim 6, wherein the first obtained data, the first other data, and the second obtained data include characteristics of a magnetic recording / reproducing device.

8. 8. The data processing apparatus according to claim 7, wherein D1 is 1.

9. one or more acquisition units; one or more processing units; the one or more acquisition units are capable of acquiring first acquisition data related to the target device in a first operation; the one or more processing units are capable of generating a first machine learning model based on first generated data based on the first acquired data and first other data; the first other data includes data related to a device similar to the target device, the first other data includes a first other feature matrix of Np rows and D1 columns and a first other label of Np rows, where Np is an integer equal to or greater than 2 and D1 is an integer equal to or greater than 1; the first acquired data includes a first feature matrix of N1 rows and D1 columns and a first acquired label of N1 rows, where N1 is an integer equal to or greater than 2 and is smaller than Np; the first generated data includes a first generator matrix of (Np+N1) rows and (3×D1) columns and a first generated label of (Np+N1) rows; the first generator matrix includes first matrix data, second matrix data, and third matrix data; an element of the first matrix data includes a combination of the first other feature amount matrix and the first feature amount matrix in a row direction; the elements of the second matrix data include a combination in the row direction of a matrix of elements of 0 in Np rows and D1 columns and the first feature amount matrix, the elements of the third matrix data include a combination in the row direction of the first other feature amount matrix and a matrix of elements of 0 with N1 rows and D1 columns, a component of the first generated label includes a combination of the first other label and the first obtained label in the row direction; A data processing system, wherein (Np+N1) / D1 is 250 or greater.

10. causing the processing unit to perform a first operation; In the first operation, the processing unit is capable of generating a first machine learning model based on first generated data that is based on first acquired data and first other data; The first acquired data relates to a target device, the first other data includes data related to a device similar to the target device, the first other data includes a first other feature matrix of Np rows and D1 columns and a first other label of Np rows, where Np is an integer equal to or greater than 2 and D1 is an integer equal to or greater than 1; the first acquired data includes a first feature matrix of N1 rows and D1 columns and a first acquired label of N1 rows, where N1 is an integer equal to or greater than 2 and is smaller than Np; the first generated data includes a first generator matrix of (Np+N1) rows and (3×D1) columns and a first generated label of (Np+N1) rows; the first generator matrix includes first matrix data, second matrix data, and third matrix data; an element of the first matrix data includes a combination of the first other feature amount matrix and the first feature amount matrix in a row direction; the elements of the second matrix data include a combination in the row direction of a matrix of elements of 0 in Np rows and D1 columns and the first feature amount matrix, the elements of the third matrix data include a combination in the row direction of the first other feature amount matrix and a matrix of elements of 0 with N1 rows and D1 columns, a component of the first generated label includes a combination of the first other label and the first obtained label in the row direction; A data processing method, wherein (Np+N1) / D1 is 250 or greater.

Images