# US 2018/0322383 A1 · Storage controller accelaration for neural network training and inference

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

- Number: US 2018/0322383 A1
- Document ID: US20180322383A1
- Status: Published application
- Kind: A1
- Application number: 15/584,136
- Filed: 2017-05-02
- Published: 2018-11-08
- Earliest claimed: 2017-05-02
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Minwei Feng; Yufei Ren; Yandong Wang; Li Zhang; Wei Zhang
- Web page: https://patentgrep.com/patent/US20180322383A1

## CPC

- G06N 3/063: using electronic means
- G06N 3/04: Architecture, e.g. interconnection topology
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 3/09: Supervised learning
- G06N 3/098: Distributed learning, e.g. federated learning
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A storage controller of a machine receives training data associated with a neural network model. The neural network model includes a plurality of layers, and the machine further including at least one graphics processing unit. The storage controller trains at least one layer of the plurality of layers of the neural network model using the training data to generate processed training data. A size of the processed data is less than a size of the training data. Training of the at least one layer includes adjusting one or more weights of the at least one layer using the training data. The storage controller sends the processed training data to at least one graphics processing unit of the machine. The at least one graphics processing unit is configured to store the processed training data and train one or more remaining layers of the plurality of layers using the processed training data.

## Claims

### Claim 1

A method comprising:
receiving, by a storage controller of a machine, training data associated with a neural network model, the neural network model including a plurality of layers, and the machine further including at least one graphics processing unit;
training, by the storage controller, at least one layer of the plurality of layers of the neural network model using the training data to generate processed training data, wherein a size of the processed data is less than a size of the training data, and wherein training of the at least one layer includes adjusting one or more weights of the at least one layer using the training data; and
sending, by the storage controller, the processed training data to at least one graphics processing unit of the machine, the at least one graphics processing unit configured to store the processed training data and train one or more remaining layers of the plurality of layers using the processed training data.

### Claim 2 (depends on 1)

The method of claim 1, wherein the training comprises performing a convolution operation on the training data, wherein the convolution operation comprises convolving the training data with the one or more weights of the at least one layer.

### Claim 3 (depends on 1)

The method of claim 1, further comprising receiving at least one gradient computed by the at least one graphics processing unit resulting from the training of the remaining layers.

### Claim 4 (depends on 4)

The method of claim 4, further comprising refining training of the at least one layer using the at least one gradient.

### Claim 5 (depends on 1)

The method of claim 1, wherein the training data comprises image data.

### Claim 6 (depends on 1)

The method of claim 1, wherein the storage controller comprises a disk controller.

### Claim 7 (depends on 1)

The method of claim 1, wherein the storage controller comprises a field-programmable gate array (FPGA).

### Claim 8

A computer usable program product comprising one or more computer-readable storage devices, and program instructions stored on at least one of the one or more storage devices, the stored program instructions comprising:
program instructions to receive, by a storage controller of a machine, training data associated with a neural network model, the neural network model including a plurality of layers, and the machine further including at least one graphics processing unit;
program instructions to train, by the storage controller, at least one layer of the plurality of layers of the neural network model using the training data to generate processed training data, wherein a size of the processed data is less than a size of the training data, and wherein training of the at least one layer includes adjusting one or more weights of the at least one layer using the training data; and
program instructions to send, by the storage controller, the processed training data to at least one graphics processing unit of the machine, the at least one graphics processing unit configured to store the processed training data and train one or more remaining layers of the plurality of layers using the processed training data.

### Claim 9 (depends on 8)

The computer usable program product of claim 8, wherein the training comprises performing a convolution operation on the training data, wherein the convolution operation comprises convolving the training data with the one or more weights of the at least one layer.

### Claim 10 (depends on 8)

The computer usable program product of claim 8, further comprising program instructions to receive at least one gradient computed by the at least one graphics processing unit resulting from the training of the remaining layers.

### Claim 11 (depends on 10)

The computer usable program product of claim 10, further comprising program instructions to refine training of the at least one layer using the at least one gradient.

### Claim 12 (depends on 8)

The computer usable program product of claim 8, wherein the training data comprises image data.

### Claim 13 (depends on 8)

The computer usable program product of claim 8, wherein the storage controller comprises a disk controller.

### Claim 14 (depends on 8)

The computer usable program product of claim 8, wherein the storage controller comprises a field-programmable gate array (FPGA).

### Claim 15 (depends on 8)

The computer usable program product of claim 8, wherein the computer usable code is stored in a computer readable storage device in a data processing system, and wherein the computer usable code is transferred over a network from a remote data processing system.

### Claim 16 (depends on 8)

The computer usable program product of claim 8, wherein the computer usable code is stored in a computer readable storage device in a server data processing system, and wherein the computer usable code is downloaded over a network to a remote data processing system for use in a computer readable storage device associated with the remote data processing system.

### Claim 17

A computer system comprising one or more processors, one or more computer-readable memories, and one or more computer-readable storage devices, and program instructions stored on at least one of the one or more storage devices for execution by at least one of the one or more processors via at least one of the one or more memories, the stored program instructions comprising:
program instructions to receive, by a storage controller of a machine, training data associated with a neural network model, the neural network model including a plurality of layers, and the machine further including at least one graphics processing unit;
program instructions to train, by the storage controller, at least one layer of the plurality of layers of the neural network model using the training data to generate processed training data, wherein a size of the processed data is less than a size of the training data, and wherein training of the at least one layer includes adjusting one or more weights of the at least one layer using the training data; and
program instructions to send, by the storage controller, the processed training data to at least one graphics processing unit of the machine, the at least one graphics processing unit configured to store the processed training data and train one or more remaining layers of the plurality of layers using the processed training data.

### Claim 18 (depends on 17)

The computer system of claim 17, wherein the training comprises performing a convolution operation on the training data, wherein the convolution operation comprises convolving the training data with the one or more weights of the at least one layer.

### Claim 19 (depends on 17)

The computer system of claim 17, further comprising program instructions to receive at least one gradient computed by the at least one graphics processing unit resulting from the training of the remaining layers.

### Claim 20 (depends on 19)

The computer system of claim 19, further comprising program instructions to refine training of the at least one layer using the at least one gradient.
