Magnetic reproducing processing device, magnetic recording and reproducing device, and magnetic reproducing method

Abstract

To provide a magnetic reproducing processing device, a magnetic recording / reproducing device, and a magnetic reproducing method capable of suppressing errors.SOLUTION: According to an embodiment, a magnetic reproducing processing device includes a decoder. The decoder includes a convolution layer including a plurality of filters and an attention layer capable of deriving a degree of contribution regarding the plurality of filters. The decoder can output a decoding result obtained by integrating the result of processing the input signal with the plurality of filters according to the degree of contribution.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

[0002] For example, fewer errors are desired in magnetic regeneration processing equipment. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] U.S. Patent No. 10,804,938 Summary of the Invention [Problem to be solved by the invention]

[0004] The embodiments of the present invention provide a magnetic reproducing processing device, a magnetic recording and reproducing device, and a magnetic reproducing method that can suppress errors. [Means for solving the problem]

[0005] According to an embodiment of the present invention, a magnetic reproducing processing device includes a decoder including a convolution layer including a plurality of filters and an attention layer capable of deriving contributions of the plurality of filters, and the decoder is capable of outputting a decoding result obtained by integrating results of processing an input signal by the plurality of filters according to the contributions. [Brief explanation of the drawings]

[0006] [Figure 1] FIG. 1 is a schematic diagram illustrating a magnetic reproducing processing device and a magnetic recording and reproducing device according to the first embodiment. [Figure 2] FIG. 2 is a schematic diagram illustrating the magnetic reproducing processing device and the magnetic recording and reproducing device according to the first embodiment. [Figure 3]FIG. 3 is a schematic diagram illustrating the magnetic reproducing processing device and the magnetic recording and reproducing device according to the first embodiment. [Figure 4] FIG. 4 is a schematic diagram illustrating the magnetic reproducing processing device and the magnetic recording and reproducing device according to the first embodiment. [Figure 5] FIG. 5 is a graph illustrating the characteristics of the magnetic regeneration processing device. [Figure 6] FIG. 6 is a schematic view illustrating a magnetic regeneration 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, and the relationship between the thickness and width of each part, the size ratio between parts, etc. are not necessarily the same as those in reality. Even when the same part is shown, the dimensions and ratios may be different depending on the drawing. In this specification and in each drawing, elements similar to those previously described 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 magnetic reproducing processing device and a magnetic recording and reproducing device according to the first embodiment. 1, the magnetic recording and reproducing device 210 according to the embodiment includes the magnetic reproducing processing device 110 according to the embodiment. The magnetic recording and reproducing device 210 may further include a magnetic recording and reproducing unit 80D.

[0009] The magnetic recording and reproducing unit 80D includes a magnetic recording medium 80. The magnetic recording medium 80 may include, for example, a magnetic disk (HDD: Hard Disk Drive). The magnetic recording and reproducing unit 80D may include, for example, an SSD (Solid State Drive). The magnetic recording and reproducing unit 80D may include, for example, a recording and reproducing unit 80R. The recording and reproducing unit 80R may include, for example, a magnetic head 80H. The recording and reproducing unit 80R (magnetic head 80H) records information on the magnetic recording medium 80. The recording and reproducing unit 80R (magnetic head 80H) reproduces the information recorded on the magnetic recording medium 80. A reproduction signal Sig-r is obtained from the recording and reproducing unit 80R.

[0010] The reproduction signal Sig-r obtained by the magnetic recording and reproducing unit 80D (recording and reproducing unit 80R) is supplied to the magnetic reproduction processing device 110. The reproduction signal Sig-r is processed (decoded) by the magnetic reproduction processing device 110. The result of the processing (decoded) by the magnetic reproduction processing device 110 is output from the magnetic reproduction processing device 110 as a decoded signal Sig-c. The decoded signal Sig-c is, for example, a binary signal "1, 0".

[0011] The magnetic reproduction processing device 110 includes a decoder 71. The magnetic reproduction processing device 110 includes, for example, an input interface 78. The reproduction signal Sig-r or a signal based on the reproduction signal Sig-r is supplied to the decoder 71 via the input interface 78. An input signal Sig-i including the reproduction signal Sig-r is supplied to the decoder 71. The input signal Sig-i (for example, the reproduction signal Sig-r) may be a continuous or discontinuous signal in a time series.

[0012] 1, the decoder 71 includes a convolutional layer 10 and an attention layer 15. As described below, the decoder 71 may include an input layer 10I, a connection layer 20, and an output layer 10O.

[0013] The convolutional layer 10 includes a plurality of filters 11. The attention layer 15 can derive contributions CR for the plurality of filters 11.

[0014] The decoder 71 can output a decoding result RR1 obtained by integrating the results of processing the input signal Sig-i by a plurality of filters 11 according to the contribution degree CR, thereby enabling decoding with higher accuracy.

[0015] For example, a first reference example may be considered in which Viterbi decoding is performed for PRML (Partial Response Maximum Likelihood). As will be described later, the decoding accuracy may be insufficient in the first reference example. On the other hand, a second reference example may be considered in which the input signal Sig-i is processed by one filter in the decoder 71. In the second reference example, one filter is used that properly processes the entire input signal Sig-i. In the second reference example, the decoding accuracy may be insufficient depending on the state of the input signal Sig-i.

[0016] In contrast, in the embodiment, decoding is performed using the results of processing the input signal Sig-i by multiple filters 11. This enables decoding with high accuracy. According to the embodiment, a magnetic reproducing processing device capable of suppressing errors can be provided.

[0017] As shown in FIG. 1, the magnetic reproducing processing device 110 may include a memory unit 10M. The memory unit 10M can store coefficients for a plurality of filters 11. The coefficients are, for example, "weights." The attention layer 15 can derive a contribution rate CR using the coefficients stored in the memory unit 10M. The attention layer 15 can, for example, estimate the contribution rate CR.

[0018] As shown in FIG. 1, the attention layer 15 estimates the contribution CR corresponding to the number cF of the plurality of filters 11 based on the coefficients stored in the memory unit 10M.

[0019] In an embodiment, the coefficients stored in the storage unit 10M may be determined by machine learning, for example.

[0020] The filters 11 may be optimized by, for example, machine learning. The decoder 71 may include, for example, a neural network (NN) structure. For example, the attention layer 15 may include a neural network (NN). The convolution layer 10 may include, for example, a convolutional neural network (CNN).

[0021] 1, the decoder 71 may further include an input layer 10I. An input signal Sig-i is input to the input layer 10I. The input signal Sig-i input to the input layer 10I is supplied to the convolutional layer 10 and the attention layer 15.

[0022] As shown in FIG. 1 , the decoder 71 may further include a combination layer 20. The combination layer 20 combines the processing result from the convolution layer 10 and the processing result from the attention layer 15. For example, the results of processing the input signal Sig-i by the multiple filters 11 are integrated according to the contribution CR. For example, the product of the results of processing the input signal Sig-i by the multiple filters 11 and the contribution CR is calculated. The sum of the calculated products is calculated. The processing result from the combination layer 20 (decoded result RR1) may be output. The decoding result RR1 may include likelihood information.

[0023] 1, the processing result of the combined layer 20 (decoding result RR1) may be supplied to another layer 25. In the other layer 25, for example, at least some of the processing results may be combined. The processing result of the other layer 25 may be supplied to the output layer 100.

[0024] 1, the decoder 71 may further include an output layer 100. The output layer 100 can output output information LH1 based on the decoding result RR1. The output information LH1 includes, for example, a likelihood proportionality.

[0025] In an embodiment, the contribution CR estimated by the attention layer 15 may vary depending on the state of the input signal Sig-i. The state of the input signal Sig-i may be, for example, the signal waveform. The state of the input signal Sig-i may be, for example, the signal length.

[0026] For example, the input signal Sig-i (reproduced signal Sig-r) is N k It includes a playback signal of T. "T" is the smallest recording unit (smallest recording period) for playback (and recording). "k" is an integer equal to or greater than 1. For example, the playback signal Sig-r includes signals such as "1T", "2T", ..., "10T", ...

[0027] For example, the input signal Sig-i is N i The playback signal of T and N j and a reproduced signal of T. "i" is an integer equal to or greater than 1. "j" is an integer equal to or greater than 1. "j" is different from "i". For example, N i The contribution CR of T to the playback signal is N j It is different from the contribution CR of T to the reproduced signal. For example, N i The coefficient for the read signal of T is N j This is different from the coefficient for the reproduced signal of T.

[0028] As shown in FIG. 1, the magnetic reproducing processing device 110 may further include an error correction decoder 72. As described above, the decoder 71 may further include an output layer 100. The output layer 100 can output output information LH1 based on the decoding result RR1. The output information LH1 can be supplied to the error correction decoder 72. Errors are corrected in the error correction decoder 72. The error correction decoder 72 may include, for example, an LDPC (Low-Density Parity-Check Codes) decoder.

[0029] As shown in Figure 1, the output of the error correction decoder 72 can be fed as part of the input signal Sig-i to the decoder 71. The iterative process results in better error suppression.

[0030] In the embodiment, the number of input nodes 10N in the input layer 10I may be, for example, 5 or more and 300,000 or less.

[0031] In the embodiment, the number of the plurality of filters 11 may be, for example, 2 or more and 1000 or less.

[0032] As shown in FIG. 1, the attention layer 15 may be capable of obtaining at least a portion of the input signal Sig-i in parallel with the convolutional layer 10.

[0033] FIG. 2 is a schematic diagram illustrating the magnetic reproducing processing device and the magnetic recording and reproducing device according to the first embodiment. 2, in the magnetic recording and reproducing device 211 according to the embodiment, the magnetic reproducing processing device 111 further includes a contribution adjuster 73. Except for this, the configuration of the magnetic reproducing processing device 111 may be the same as the configuration of the magnetic reproducing processing device 110.

[0034] The contribution adjuster 73 can adjust the contributions CR of the multiple filters 11. The contribution adjuster 73 adjusts the contributions CR of at least some of the multiple filters 11 based on at least some of the processing results of the error correction decoder 72. In this example, at least some of the processing results of the error correction decoder 72 are supplied to the contribution adjuster 73. In the embodiment, the method of adjusting the contributions CR in the contribution adjuster 73 may be modified in various ways. For example, the contributions CR may be adjusted to match the characteristics of the target magnetic recording and reproducing unit 80D. More appropriate decoding can be performed.

[0035] FIG. 3 is a schematic diagram illustrating the magnetic reproducing processing device and the magnetic recording and reproducing device according to the first embodiment. 3, in the magnetic recording and reproducing device 212 according to the embodiment, the magnetic reproducing processing device 112 further includes a waveform controller (WC) 74. Except for this, the configuration of the magnetic reproducing processing device 112 may be the same as the configuration of the magnetic reproducing processing device 110.

[0036] The decoder 71 can acquire the input signal Sig-i after being adjusted by the waveform adjuster 74. For example, the waveform adjuster 74 acquires the reproduction signal Sig-r, adjusts the waveform of the reproduction signal Sig-r, and outputs it as the input signal Sig-i. The decoder 71 (e.g., the input layer 10I) may acquire the input signal Sig-i adjusted by the waveform adjuster 74. For example, the waveform is adjusted to match the characteristics of the target magnetic recording and reproducing unit 80D. More appropriate decoding is possible. The waveform adjuster 74 may include, for example, FIR (Finite Impulse Response) or the like.

[0037] FIG. 4 is a schematic diagram illustrating the magnetic reproducing processing device and the magnetic recording and reproducing device according to the first embodiment. 4, in the magnetic recording and reproducing device 213 according to the embodiment, the magnetic reproducing processing device 113 includes a plurality of decoders 71. Except for this, the configuration of the magnetic reproducing processing device 113 may be the same as the configuration of the magnetic reproducing processing device 110.

[0038] In the magnetic reproducing processing device 113, the multiple decoders 71 are capable of parallel processing. The processing results from the multiple decoders 71 are combined by a combiner 75. The output of the combiner 75 is supplied to the error correction decoder 72. In this way, for example, the combiner 75 can combine a decoding result RR1 obtained from one of the multiple decoders 71 with a decoding result RR1 obtained from another of the multiple decoders 71. For example, the combiner 75 combines output information LH1 (e.g., proportional to likelihood) obtained from one of the multiple decoders 71 with output information LH1 (e.g., proportional to likelihood) obtained from another of the multiple decoders 71. The result combined by the combiner 75 is supplied to the error correction decoder 72. The processing results from the error correction decoder 72 may be supplied to the multiple decoders 71. More appropriate decoding is possible. For example, faster processing is possible.

[0039] For example, a first learning condition for one of the multiple decoders 71 is different from a second learning condition for another of the multiple decoders 71. In one example, a first error function under the first learning condition is different from a second error function under the second learning condition. In one example, the order of the multiple learning data values ​​under the first learning condition is different from the order of the multiple learning data values ​​under the second learning condition. For example, the order of the learning data values ​​under the first learning condition is reversed from that under the second learning condition. For example, in learning, the reproduced signal Sig-r is used as learning data. The reproduced signal Sig-r is expressed as multiple signal intensity values ​​at multiple times. The learning data includes a first value to a k-th value, where "k" is an integer equal to or greater than 2. "k" corresponds to time. Under the first learning condition, learning is performed from the first value to the k-th value. Under the second learning condition, learning is performed from the k-th value to the first value. More appropriate decoding is possible by processing the learning data using multiple decoders 71 under different learning conditions.

[0040] FIG. 5 is a graph illustrating the characteristics of the magnetic regeneration processing device. FIG. 5 illustrates the characteristics of the magnetic reproducing processing devices 110 to 113. Furthermore, FIG. 5 also illustrates the characteristics of the magnetic reproducing processing device 118 of the first reference example and the characteristics of the magnetic reproducing processing device 119 of the second reference example. In the magnetic reproducing processing device 118, Viterbi decoding is performed using a PR filter. This corresponds to the fact that the number of filters in the decoding of the first reference example is one. In the magnetic reproducing processing device 119 of the second reference example, the reproduced signal is processed by one filter. In the first reference example, the decoder 71 is replaced with a Viterbi decoder. In the second reference example, the decoder 71 is processed by one filter. Except for the above, the configurations of the magnetic reproducing processing devices 118 and 119 are the same as the configuration of the magnetic reproducing processing device 110. The horizontal axis of FIG. 5 represents the number of repetitions nx of the process. The number of repetitions nx is displayed in a normalized form. The vertical axis of FIG. 5 represents the BER (Bit Error Rate).

[0041] In the example of FIG. 5, the number of nodes in the input layer 10I is 11, and the number of channels is 1. The number of filters 11 included in the convolutional layer 10 is 15. In other layers 25 (e.g., fully connected layers), the number of layers is 5, and the number of nodes is 10. The number of output nodes in the output layer 10O is 1. In this example, the attention layer 15 is a fully connected layer, and in that fully connected layer, the number of layers is 5, the number of input nodes is 11, and the number of output nodes is 15. The output of the attention layer 15 is a softmax output.

[0042] 5, in the magnetic regeneration processing devices 110 to 113 according to the embodiment, a lower BER is obtained than the BER in the magnetic regeneration processing device 118 of the first reference example and the magnetic regeneration processing device 119 of the second reference example. The BER in the magnetic regeneration processing device 111 is lower than the BER in the magnetic regeneration processing device 110. The BER in the magnetic regeneration processing device 112 is lower than the BER in the magnetic regeneration processing device 111. The BER in the magnetic regeneration processing device 113 is lower than the BER in the magnetic regeneration processing device 112.

[0043] The magnetic regeneration processing device according to the embodiment may include a computer. FIG. 6 is a schematic view illustrating a magnetic regeneration processing device according to the embodiment. 6, a magnetic reproducing processing device according to an embodiment (e.g., the magnetic reproducing processing device 110) may include a processing circuit 76p, a memory circuit 76m, and an interface circuit 76f. The processing circuit 76p is, for example, an electric circuit. The memory circuit 76m may include, for example, at least one of a read-only memory (ROM) and a random access memory (RAM). For example, a part of the magnetic recording and reproducing unit 80D may be used as the memory circuit 76m.

[0044] A magnetic regeneration processing device according to an embodiment (e.g., the magnetic regeneration processing device 110) may include a display device 76d and an input device 76i. The display device 76d may include various types of displays. The input device 76i may include, for example, a device having an operation function (e.g., a keyboard, a mouse, a touch-type input panel, or a voice recognition input device).

[0045] The multiple elements included in the magnetic regeneration processing device according to the embodiment (e.g., the magnetic regeneration processing device 110) can communicate with each other by at least one of wireless and wired methods. The multiple elements included in the magnetic regeneration processing device 110 may be installed in different locations. For example, a general-purpose computer may be used as the magnetic regeneration processing device 110. For example, multiple computers connected to each other may be used as the magnetic regeneration processing device 110. A dedicated circuit may be used as at least a part of the magnetic regeneration processing device 110. For example, multiple circuits connected to each other may be used as the magnetic regeneration processing device 110.

[0046] The embodiment may include a program that causes a computer (magnetic reproducing processing device 110) to perform the above-described operations. The embodiment may also include a storage medium on which the above-described program is stored.

[0047] (Second embodiment) The second embodiment relates to a magnetic reproducing method. The magnetic reproducing method according to the embodiment is a method that uses the magnetic reproducing processing device (for example, 110 to 113) according to the first embodiment and variations thereof. A magnetic reproducing method that can suppress errors can be provided.

[0048] The embodiment may include the following configurations (e.g., technical solutions). (Configuration 1) a decoder; The decoder a convolutional layer including multiple filters; an attention layer capable of deriving contributions of the plurality of filters; Including, The decoder is capable of outputting a decoding result obtained by integrating results of processing an input signal by the plurality of filters according to the degree of contribution.

[0049] (Configuration 2) a storage unit capable of storing coefficients related to the plurality of filters; 2. The magnetic reproducing processing device according to claim 1, wherein the attention layer is capable of deriving the contribution using the coefficient.

[0050] (Configuration 3) 3. The magnetic reproducing processing device according to configuration 1 or 2, wherein the decoding result includes likelihood information.

[0051] (Configuration 4) 4. The magnetic reproducing processing device according to any one of configurations 1 to 3, wherein the input signal includes a reproduced signal obtained from a magnetic recording and reproducing unit.

[0052] (Configuration 5) The input signal is N i The playback signal of T and N j a playback signal of T, The T is the minimum recording unit for playback, The i is an integer of 1 or more, The j is an integer of 1 or more, The j is different from the i, N i The contribution of T to the reproduced signal is j 5. The magnetic reproducing processing device according to any one of configurations 1 to 4, wherein the contribution of T to the reproduced signal is different from that of T.

[0053] (Configuration 6) the decoder further includes an output layer; the output layer is capable of outputting output information based on the decoding result; 6. The magnetic reproducing processing device according to any one of configurations 1 to 5, wherein the output information includes likelihood proportion.

[0054] (Configuration 7) further comprising an error correction decoder; the decoder further includes an output layer; the output layer is capable of outputting output information based on the decoding result; 6. The magnetic reproducing processing device according to any one of configurations 1 to 5, wherein the output information can be supplied to the error correction decoder.

[0055] (Configuration 8) 8. The magnetic reproducing processing device of claim 7, wherein an output of the error correction decoder can be supplied to the decoder as part of the input signal.

[0056] (Configuration 9) 9. The magnetic regeneration processing device according to any one of configurations 1 to 8, wherein the number of the plurality of filters is 2 or more and 1000 or less.

[0057] (Configuration 10) 10. The magnetic reproducing processing device according to any one of configurations 1 to 9, wherein the attention layer is capable of acquiring at least a part of the input signal in parallel with the convolution layer.

[0058] (Configuration 11) the decoder further includes an input layer; The input signal is input to the input layer, The input signal input to the input layer is supplied to the convolution layer and the attention layer, 11. The magnetic reproducing processing device according to any one of configurations 1 to 10, wherein the number of input nodes in the input layer is 5 or more and 300,000 or less.

[0059] (Configuration 12) Further comprising a waveform adjuster; 11. The magnetic reproducing processing device according to any one of configurations 1 to 10, wherein the decoder is capable of acquiring the input signal after being adjusted by the waveform adjuster.

[0060] (Configuration 13) Further comprising a contribution adjuster; 13. The magnetic reproducing processing device according to any one of configurations 1 to 12, wherein the contribution adjuster is capable of adjusting the contributions of the plurality of filters.

[0061] (Configuration 14) further comprising a coupler; a plurality of said decoders are provided; A magnetic reproducing processing device described in any one of configurations 1 to 13, wherein the combiner is capable of combining the decoding result obtained from the plurality of decoders with the decoding result obtained from another one of the plurality of decoders.

[0062] (Configuration 15) a plurality of said decoders are provided; 14. The magnetic reproducing processing device according to any one of configurations 1 to 13, wherein a first learning condition in the plurality of decoders is different from a second learning condition in another one of the plurality of decoders.

[0063] (Configuration 16) 16. The magnetic regeneration processing device according to configuration 15, wherein a first error function under the first learning condition is different from a second error function under the second learning condition.

[0064] (Configuration 17) 16. The magnetic regeneration processing device according to configuration 15, wherein an arrangement of the plurality of learning data values ​​under the first learning condition is different from an arrangement of the plurality of learning data values ​​under the second learning condition.

[0065] (Configuration 18) 16. The magnetic reproducing processing device according to any one of configurations 1 to 15, wherein the decoder includes a neural network structure.

[0066] (Configuration 19) The magnetic regeneration processing device according to configuration 4, the magnetic recording and reproducing unit; A magnetic recording and reproducing device comprising:

[0067] (Configuration 20) 19. A magnetic regeneration method using the magnetic regeneration processing device according to any one of configurations 1 to 18.

[0068] According to the embodiments, it is possible to provide a magnetic reproducing processing device, a magnetic recording and reproducing device, and a magnetic reproducing method that can suppress errors.

[0069] 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 magnetic reproducing processing device, such as the decoder, error correction decoder, waveform adjuster, and contribution adjuster, 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 known ranges.

[0070] 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.

[0071] All magnetic reproducing processing devices, magnetic recording and reproducing devices, and magnetic reproducing methods that can be implemented by a person skilled in the art by making appropriate design modifications based on the magnetic reproducing processing devices, magnetic recording and reproducing devices, and magnetic reproducing methods 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.

[0072] 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.

[0073] 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]

[0074] 10: convolution layer, 10I: input layer, 10M: memory unit, 10N: input node, 10O: output layer, 11: filter, 15: attention layer, 20: connection layer, 25: other layers, 71: decoder, 72: error correction decoder, 73: contribution adjuster, 74: waveform adjuster, 75: combiner, 76d: display device, 76f: interface circuit, 76i: input device, 76m: memory circuit, 76p: processing circuit, 78: input interface, 80: magnetic recording medium, 80D: magnetic recording and reproducing unit, 80H: magnetic head, 80R: recording and reproducing unit, 110-113, 118, 119: magnetic reproducing processing device, 210-213: magnetic recording and reproducing device, CR: contribution, LH1: output information, RR1: decoding result, Sig-c: Decoded signal, Sig-i: Input signal, Sig-r: Regenerated signal, cF: Filter number, nx: Number of repetitions

Claims

1. a decoder; The decoder a convolutional layer including multiple filters; an attention layer capable of deriving contributions of the plurality of filters; Including, A time-series input signal is input to the convolution layer and the attention layer; the decoder is capable of outputting a decoding result obtained by integrating results of processing the time-series input signal by the plurality of filters in accordance with the contribution degree derived based on the time-series input signal; the input signal includes a reproduced signal of N i T and a reproduced signal of N j T; The T is the minimum recording unit for playback, The i is an integer of 1 or more, The j is an integer of 1 or more, The j is different from the i, The magnetic reproducing processing device, wherein the contribution of the N i T to the reproduced signal is different from the contribution of the N j T to the reproduced signal.

2. a storage unit capable of storing coefficients related to the plurality of filters; The magnetic reproducing processing device according to claim 1 , wherein the attention layer is capable of deriving the contribution using the coefficient.

3. 2. The magnetic reproducing processing device according to claim 1, wherein the input signal includes a reproduced signal obtained from a magnetic recording and reproducing unit.

4. the decoder further includes an output layer; the output layer is capable of outputting output information based on the decoding result; The magnetic reproducing processing device according to claim 1 , wherein the output information includes a likelihood ratio.

5. further comprising an error correction decoder; the decoder further includes an output layer; the output layer is capable of outputting output information based on the decoding result; 2. The magnetic reproducing processing device of claim 1, wherein the output information is capable of being supplied to the error correction decoder.

6. 6. The magnetic reproducing processing apparatus of claim 5, wherein an output of said error correction decoder can be supplied to said decoder as part of said input signal.

7. Further comprising a waveform adjuster; The magnetic reproducing processing device according to claim 1 , wherein the decoder is capable of obtaining the input signal after being adjusted by the waveform adjuster.

8. Further comprising a contribution adjuster; The magnetic regeneration processing device of claim 1 , wherein the contribution adjuster is capable of adjusting the contributions of the plurality of filters.

9. a plurality of said decoders are provided; 2. The magnetic reproducing processing device according to claim 1, wherein a first learning condition of one of the plurality of decoders is different from a second learning condition of another of the plurality of decoders.

10. The magnetic regeneration processing device according to claim 3; the magnetic recording and reproducing unit; A magnetic recording and reproducing device comprising:

11. A magnetic regeneration method using the magnetic regeneration processing device according to any one of claims 1 to 9.

Images