# US 2019/0318260 A1 · Recording medium with machine learning program recorded therein, machine learning method, and information processing apparatus

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2019/0318260 A1
- Document ID: US20190318260A1
- Status: Published application
- Kind: A1
- Application number: 16/364,583
- Filed: 2019-03-26
- Published: 2019-10-17
- Earliest claimed: 2018-04-12
- Assignees: Fujitsu Limited
- Applicants: Fujitsu Limited
- Inventors: Suguru YASUTOMI; TAKASHI KATOH; Kento UEMURA
- Web page: https://patentgrep.com/patent/US20190318260A1

## CPC

- G06N 3/08: Learning methods
- G06N 20/00: Machine learning
- G06F 17/14: Fourier, Walsh or analogous domain transformations {, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms (for correlation function computation G06F17/156; spectrum analysers G01R23/16)}
- G06F 17/15: Correlation function computation {including computation of convolution operations (arithmetic circuits for sum of products per se, e.g. multiply-accumulators G06F7/5443; digital filters, e.g. FIR, IIR, adaptive filters H03H17/00)}
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/09: Supervised learning
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A non-transitory computer-readable recording medium with a machine learning program recorded therein for enabling a computer to perform processing includes: generating augmented data by data-augmenting at least some data of training data or at least some data of data input to a convolutional layer included in a learner, using a filter corresponding to a size depending on details of the processing of the convolutional layer or a filter corresponding to a size of an identification target for the learner; and learning the learner using the training data and the augmented data.

## Claims

### Claim 1

A non-transitory computer-readable recording medium with a machine learning program recorded therein for enabling a computer to perform processing, comprising:
generating augmented data by data-augmenting at least some data of training data or at least some data of data input to a convolutional layer included in a learner, using a filter corresponding to a size depending on details of the processing of the convolutional layer or a filter corresponding to a size of an identification target for the learner; and
learning the learner using the training data and the augmented data.

### Claim 2 (depends on 1)

The non-transitory computer-readable recording medium according to claim 1, wherein
the generating augmented data includes generating the augmented data by data-augmenting data of an intermediate layer of the learner using the filter.

### Claim 3 (depends on 1)

The non-transitory computer-readable recording medium according to claim 1, wherein
the generating augmented data includes generating the augmented data by data-augmenting data of an input layer of the learner using the filter.

### Claim 4 (depends on 1)

The non-transitory computer-readable recording medium according to claim 1, wherein
the generating augmented data includes generating the augmented data by Fourier-transforming the data and data-augmenting the Fourier-transformed data by eliminating frequency components higher than a peak frequency.

### Claim 5 (depends on 1)

The non-transitory computer-readable recording medium according to claim 1, wherein
the generating augmented data includes generating the augmented data by augmenting the data by adding noise to the data, the noise to achieve a degree of a blur depending on a size of a sliding window of the convolutional layer.

### Claim 6 (depends on 1)

The non-transitory computer-readable recording medium according to claim 1, wherein
the generating augmented data includes generating the augmented data by applying a parameter with the largest value of a loss function from among a plurality of parameters of the learner for which data augmentation has been successful, depending on the progress of a learning process of the learner.

### Claim 7

A machine learning method comprising:
generating, by a computer, augmented data by data-augmenting at least some data of training data or at least some data of data input to a convolutional layer included in a learner, using a filter corresponding to a size depending on details of the processing of the convolutional layer or a filter corresponding to a size of an identification target for the learner; and
learning the learner using the training data and the augmented data.

### Claim 8 (depends on 7)

The machine learning method according to claim 7, wherein
the generating augmented data includes generating the augmented data by data-augmenting data of an intermediate layer of the learner using the filter.

### Claim 9 (depends on 7)

The machine learning method according to claim 7, wherein
the generating augmented data includes generating the augmented data by data-augmenting data of an input layer of the learner using the filter.

### Claim 10 (depends on 7)

The machine learning method according to claim 7, wherein
the generating augmented data includes generating the augmented data by Fourier-transforming the data and data-augmenting the Fourier-transformed data by eliminating frequency components higher than a peak frequency.

### Claim 11 (depends on 7)

The machine learning method according to claim 7, wherein
the generating augmented data includes generating the augmented data by augmenting the data by adding noise to the data, the noise to achieve a degree of a blur depending on a size of a sliding window of the convolutional layer.

### Claim 12 (depends on 7)

The machine learning method according to claim 7, wherein
the generating augmented data includes generating the augmented data by applying a parameter with the largest value of a loss function from among a plurality of parameters of the learner for which data augmentation has been successful, depending on the progress of a learning process of the learner.

### Claim 13

An information processing apparatus comprising:
a memory; and
a processor coupled to the memory and configured to execute a processing of:
generating augmented data by data-augmenting at least some data of training data or at least some data of data input to a convolutional layer included in a learner, using a filter corresponding to a size depending on details of the processing of the convolutional layer or a filter corresponding to a size of an identification target for the learner; and
learning the learner using the training data and the augmented data.

### Claim 14 (depends on 13)

The information processing apparatus according to claim 13, wherein
the generating augmented data includes generating the augmented data by data-augmenting data of an intermediate layer of the learner using the filter.

### Claim 15 (depends on 13)

The information processing apparatus according to claim 13, wherein
the generating augmented data includes generating the augmented data by data-augmenting data of an input layer of the learner using the filter.

### Claim 16 (depends on 13)

The information processing apparatus according to claim 13, wherein
the generating augmented data includes generating the augmented data by Fourier-transforming the data and data-augmenting the Fourier-transformed data by eliminating frequency components higher than a peak frequency.

### Claim 17 (depends on 13)

The information processing apparatus according to claim 13, wherein
the generating augmented data includes generating the augmented data by augmenting the data by adding noise to the data, the noise to achieve a degree of a blur depending on a size of a sliding window of the convolutional layer.

### Claim 18 (depends on 13)

The information processing apparatus according to claim 13, wherein
the generating augmented data includes generating the augmented data by applying a parameter with the largest value of a loss function from among a plurality of parameters of the learner for which data augmentation has been successful, depending on the progress of a learning process of the learner.