# US 2026/0101053 A1 · Method, apparatus and system for encoding and decoding a tensor

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

- Number: US 2026/0101053 A1
- Document ID: US20260101053A1
- Status: Published application
- Kind: A1
- Application number: 19/114,185
- Filed: 2023-07-28
- Published: 2026-04-09
- Earliest claimed: 2022-10-13
- Assignees: Canon Kabushiki Kaisha
- Applicants: Canon Kabushiki Kaisha
- Inventors: Christopher James ROSEWARNE; Thi Hong Nhung NGUYEN
- Web page: https://patentgrep.com/patent/US20260101053A1

## CPC

- H04N 19/33: in the spatial domain
- H04N 19/463: by compressing encoding parameters before transmission
- G06N 3/02: Neural networks
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/048: Activation functions
- G06N 3/063: using electronic means
- G06N 3/08: Learning methods
- G06N 3/082: modifying the architecture, e.g. adding, deleting or silencing nodes or connections
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 3/088: Non-supervised learning, e.g. competitive learning
- G06T 9/002: {using neural networks}
- H04N 19/136: Incoming video signal characteristics or properties
- H04N 19/172: the region being a picture, frame or field
- H04N 19/42: characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation (H04N19/635 takes precedence)
- H04N 19/50: using predictive coding (H04N19/61 takes precedence)
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

An apparatus and method for encoding. The method comprises performing data compression for data related to an image, the data compression using a neural network associated with a first set of weights; determining whether a set of weights for the data compression is to be changed; and encoding first information indicating whether the set of weights is to be changed, based on the determination. If it is determined that the set of weights for the data compression is to be changed, the method comprises changing association of the neural network used in the data compression to a second set of weights from the first set of weights; encoding second information indicating the second set of weights; and encoding the data in which the data compression is performed using the neural network with the associated second set of weights.

## Claims

### Claim 1

A method of encoding, the method comprising:
performing data compression for data related to an image, the data compression using a neural network associated with a first set of weights;
determining whether a set of weights for the data compression is to be changed; and
encoding first information indicating whether the set of weights is to be changed, based on the determination,
wherein, if it is determined that the set of weights for the data compression is to be changed, the method further comprises:
changing association of the neural network used in the data compression to a second set of weights from the first set of weights;
encoding second information indicating the second set of weights; and
encoding the data in which the data compression is performed using the neural network with the associated second set of weights.

### Claim 2 (depends on 1)

The method according to claim 1, wherein the data compression includes multi-scale feature compression MSFC processing using the neural network.

### Claim 3 (depends on 1)

The method according to claim 1, wherein an input to the data compression is at least one tensor in which neural network processing for a machine task is partially performed.

### Claim 4 (depends on 1)

The method according to claim 1, wherein the second information represents a delta between the first set of weights and the second set of weights.

### Claim 5 (depends on 1)

The method according to claim 1, further comprising:
acquiring a first value for evaluating data compression using the neural network with the first set of weights; and
acquiring a second value for evaluating data compression using the neural network with the second set of weights,
wherein, whether a set of weights for the data compression is changed is determined, based on a result of comparison of the first value and the second value.

### Claim 6 (depends on 5)

The method according to claim 5, wherein the first and second values are values of MSE (Mean Square Error).

### Claim 7 (depends on 5)

The method according to claim 5, wherein the first value is acquired using a tensor input to the data compression and a tensor after the data compression using the neural network with the first set of weights.

### Claim 8

A method of decoding, the method comprising:
performing decoding processing to decode data related to an image from encoded data in which data compression is performed, the decoding performed using a neural network associated with a first set of weights; and
decoding first information indicating whether a set of weights for the decoding processing is to be changed,
wherein, if the first information indicates that the set of weights for the decoding processing is to be changed, the method further comprises:
decoding second information indicating a second set of weights; and
changing the association of the neural network used in the decoding processing to the second set of weights from the first set of weights.

### Claim 9 (depends on 8)

The method according to claim 8, wherein the decoding processing including multi-scale feature compression (MSFC) processing using the neural network.

### Claim 10 (depends on 8)

The method according to claim 8, wherein the encoded data includes data of a tensor in which neural network processing for a machine task is partially performed.

### Claim 11 (depends on 10)

The method according to claim 10,
the method further comprising performing a remaining portion of the neural network processing for a machine task.

### Claim 12 (depends on 8)

The method according to claim 8, wherein the second information represents delta between the first set of weights and the second set of weights.

### Claim 13

A non-transitory computer-readable storage medium which stores a program for executing a method of encoding, the method comprising:
performing data compression for data related to an image, the data compression using a neural network associated with a first set of weights;
determining whether a set of weights for the data compression is to be changed; and
encoding first information indicating whether the set of weights is to be changed, based on the determination,
wherein, if it is determined that the set of weights for the data compression is to be changed, the method further comprises:
changing association of the neural network used in the data compression to a second set of weights from the first set of weights;
encoding second information indicating the second set of weights; and
encoding the data in which the data compression is performed using the neural network with the associated second set of weights.

### Claim 14

An encoder configured to:
perform data compression for data related to an image, the data compression using a neural network associated with a first set of weights;
determine whether a set of weights for the data compression is to be changed; and
encode first information indicating whether the set of weights is to be changed, based on the determination,
wherein, if it is determined that the set of weights for the data compression is to be changed, the encoder is further configured to:
change association of the neural network used in the data compression to a second set of weights from the first set of weights;
encode second information indicating the second set of weights; and
encode the data in which the data compression is performed using the neural network with the associated second set of weights.

### Claim 15

A system comprising:
a memory; and
a processor, wherein the processor is configured to execute code stored on the memory for implementing a method comprising:
performing data compression for data related to an image, the data compression using a neural network associated with a first set of weights;
determining whether a set of weights for the data compression is to be changed; and
encoding first information indicating whether the set of weights is to be changed, based on the determination,
wherein, if it is determined that the set of weights for the data compression is to be changed, the method further comprises:
changing association of the neural network used in the data compression to a second set of weights from the first set of weights;
encoding second information indicating the second set of weights; and
encoding the data in which the data compression is performed using the neural network with the associated second set of weights.

### Claim 16

A non-transitory computer-readable storage medium which stores a program for executing a method of decoding, the method comprising:
performing decoding processing to decode data related to an image from encoded data in which data compression is performed, the decoding performed using a neural network associated with a first set of weights; and
decoding first information indicating whether a set of weights for the decoding processing is to be changed,
wherein, if the first information indicates that the set of weights for the decoding processing is to be changed, the method further comprises:
decoding second information indicating a second set of weights; and
changing the association of the neural network used in the decoding processing to the second set of weights from the first set of weights.

### Claim 17

A decoder configured to:
perform decoding processing to decode data related to an image from encoded data in which data compression is performed, the decoding performed using a neural network associated with a first set of weights; and
decode first information indicating whether a set of weights for the decoding processing is to be changed,
wherein, if the first information indicates that the set of weights for the decoding processing is to be changed, the decoder is configured to:
decode second information indicating a second set of weights; and
change the association of the neural network used in the decoding processing to the second set of weights from the first set of weights.

### Claim 18

A system comprising:
a memory; and
a processor, wherein the processor is configured to execute code stored on the memory for implementing a method comprising:
performing decoding processing to decode data related to an image from encoded data in which data compression is performed, the decoding performed using a neural network associated with a first set of weights; and
decoding first information indicating whether a set of weights for the decoding processing is to be changed,
wherein, if the first information indicates that the set of weights for the decoding processing is to be changed, the method further comprises:
decoding second information indicating a second set of weights; and
changing the association of the neural network used in the decoding processing to the second set of weights from the first set of weights.
