# US 10,482,375 B2 · Deep graph representation learning

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

- Number: US 10,482,375 B2
- Document ID: US10482375B2
- Status: Granted
- Kind: B2
- Application number: 15/802,302
- Filed: 2017-11-02
- Published: 2019-05-02
- Granted: 2019-11-19
- Earliest claimed: 2017-11-02
- Assignees: Palo Alto Research Incorporated
- Applicants: Palo Alto Research Center Incorporated
- Inventors: Ryan Rossi; Rong Zhou
- Web page: https://patentgrep.com/patent/US10482375B2

## CPC

- G06N 20/00: Machine learning
- G06F 16/9024: {Graphs; Linked lists (G06F16/9027 takes precedence)}
- G06F 18/213: Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06N 3/0495: Quantised networks; Sparse networks; Compressed networks
- G06N 3/096: Transfer learning
- G06N 5/022: Knowledge engineering; Knowledge acquisition
- G06V 10/426: Graphical representations
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A method of deep graph representation learning includes: calculating a plurality of base features from a graph and adding the plurality of base features to a feature matrix. The method further includes generating, by a processing device, a current feature layer from the feature matrix and a set of relational feature operators, wherein the current feature layer corresponds to a set of current features, evaluating feature pairs associated with the current feature layer, and selecting a subset of features from the set of current features based on the evaluated feature pairs. The method further includes adding the subset of features to the feature matrix to generate an updated feature matrix.

## Claims

### Claim 1

A method of deep graph representation learning, the method comprising:
calculating a plurality of base features from a graph;
adding the plurality of base features to a feature matrix;
(i) generating, by a processing device, a current feature layer from the feature matrix and a set of relational feature operators, wherein the current feature layer corresponds to a set of current features;
(ii) evaluating feature pairs associated with the current feature layer;
(iii) selecting a subset of features from the set of current features based on the evaluated feature pain; and
(iv) adding the subset of features to the feature matrix to generate an updated feature matrix.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
incrementing the current feature layer to generate a new current feature layer; and
repeating steps (i) through (iv) for the new current feature layer until no new features emerge or a max number of layers is reached.

### Claim 3 (depends on 1)

The method of claim 1, further comprising:
transforming the plurality of base features after generating a current feature layer; and
adding the plurality of transformed base features to the updated feature matrix to generate a new updated feature matrix.

### Claim 4 (depends on 1)

The method of claim 1, further comprising: transforming the set of current features after generating a current feature layer.

### Claim 5 (depends on 1)

The method of claim 1, wherein a plurality of features in the feature matrix are transfer learning features.

### Claim 6 (depends on 1)

The method of claim 1, wherein the graph is one of a directed graph, an undirected graph, a bipartite graph, or a weighted graph.

### Claim 7 (depends on 6)

The method of claim 6, wherein selecting the subset of features further comprises: applying a set of relational feature operators to each feature of a previous feature layer.

### Claim 8

A system comprising:
a memory; and
a processing device, operatively coupled to the memory, to:
calculate a plurality of base features from a graph;
add the plurality of base features to a feature matrix;
(i) generate a current feature layer from the feature matrix and a set of relational feature operators, wherein the current feature layer corresponds to a set of current features;
(ii) evaluate feature pairs associated with the current feature layer;
(iii) select a subset of features from the set of current features based on the evaluated feature pairs; and
(iv) add the subset of features to the feature matrix to generate an updated feature matrix.

### Claim 9 (depends on 8)

The system of claim 8, wherein the processing device is further to:
increment the current feature layer to generate a new current feature layer, and
repeat steps (i) through (iv) for the new current feature layer until no new features emerge or a max number of layers is reached.

### Claim 10 (depends on 8)

The system of claim 8, wherein the processing device is further to:
transform the plurality of base features after generating a current feature layer; and
add the plurality of transformed base features to the updated feature matrix to generate a new updated feature matrix.

### Claim 11 (depends on 8)

The system of claim 8, wherein the processing device is further to: transform the set of current features after generating a current feature layer.

### Claim 12 (depends on 8)

The system of claim 8, wherein the processing device is one or more graphics processing units of one or more servers.

### Claim 13 (depends on 8)

The system of claim 8, wherein the graph is one of a directed graph, an undirected graph, a bipartite graph, or a weighted graph.

### Claim 14 (depends on 13)

The system of claim 13, wherein to select the subset of features the processing device is to: apply a set of relational feature operators to each feature of a previous feature layer.

### Claim 15

A non-transitory computer-readable storage medium having instructions stored thereon that, when executed by a processing device, cause the processing device to:
calculate a plurality of base features from a graph;
add the plurality of base features to a feature matrix;
(i) generate, by the processing device, a current feature layer from the feature matrix and a set of relational feature operators, wherein the current feature layer corresponds to a set of current features;
(ii) evaluate feature pairs associated with the current feature layer;
(iii) select a subset of features from the set of current features based on the evaluated feature pairs; and
(iv) add the subset of features to the feature matrix to generate an updated feature matrix.

### Claim 16 (depends on 15)

The non-transitory computer-readable storage medium of claim 15, wherein the processing device is further to:
increment the current feature layer to generate a new current feature layer; and
repeat steps (i) through (iv) for the new current feature layer until no new features emerge or a max number of layers is reached.

### Claim 17 (depends on 15)

The non-transitory computer-readable storage medium of claim 15, wherein the processing device is further to:
transform the plurality of base features after generating a current feature layer; and
add the plurality of transformed base features to the updated feature matrix to generate a new updated feature matrix.

### Claim 18 (depends on 15)

The non-transitory computer-readable storage medium of claim 15, wherein the processing device is further to: transform the set of current features after generating a current feature layer.

### Claim 19 (depends on 15)

The non-transitory computer-readable storage medium of claim 15, wherein a plurality of features in the feature matrix are transfer learning features.

### Claim 20 (depends on 15)

The non-transitory computer-readable storage medium of claim 15, wherein the graph is one of a directed graph, an undirected graph, a bipartite graph, or a weighted graph.