# US 2019/0171765 A1 · Adaptive clustering of media content from multiple different domains

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

- Number: US 2019/0171765 A1
- Document ID: US20190171765A1
- Status: Published application
- Kind: A1
- Application number: 15/829,155
- Filed: 2017-12-01
- Published: 2019-06-06
- Earliest claimed: 2017-12-01
- Assignees: AT&T Intellectual Property I, L.P.
- Applicants: AT&T Intellectual Property I, L.P.
- Inventors: Raghuraman Gopalan
- Web page: https://patentgrep.com/patent/US20190171765A1

## CPC

- G06F 16/9535: Search customisation based on user profiles and personalisation
- G06F 16/43: Querying
- G06F 16/55: Clustering; Classification
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

In one example, the present disclosure describes a device, computer-readable medium, and method for adaptively clustering media content from multiple different domains in the presence of domain shift. For instance, in one example, a plurality of data content items is acquired from a plurality of different domains, wherein at least some data content items of the plurality of data content items are unlabeled. The plurality of data content items is encoded with a feature representing a domain shift variation that is assumed to be present in the plurality of data content items, wherein the domain shift variation comprises variation in a characteristic of the plurality of data content items. The plurality of data content items is clustered into a predefined number of content categories subsequent to the encoding.

## Claims

### Claim 1

A method, comprising:
acquiring a plurality of data content items from a plurality of different domains, wherein at least some data content items of the plurality of data content items are unlabeled;
encoding the plurality of data content items with a feature representing a domain shift variation that is assumed to be present in the plurality of data content items, wherein the domain shift variation comprises variation in a characteristic of the plurality of data content items; and
clustering the plurality of data content items into a predefined number of content categories subsequent to the encoding.

### Claim 2 (depends on 1)

The method of claim 1, wherein a number of the plurality of different domains is unknown.

### Claim 3 (depends on 1)

The method of claim 1, wherein a type of the domain shift variation is unknown.

### Claim 4 (depends on 1)

The method of claim 1, wherein the plurality of data content items comprises a plurality of images, and the characteristic comprises illumination.

### Claim 5 (depends on 1)

The method of claim 1, wherein the plurality of data content items comprises a plurality of images, and the characteristic comprises viewpoint.

### Claim 6 (depends on 1)

The method of claim 1, wherein the plurality of data content items comprises a plurality of images, and the characteristic comprises an affine transformation.

### Claim 7 (depends on 1)

The method of claim 1, wherein the plurality of data content items comprises a plurality of images, and the characteristic comprises resolution.

### Claim 8 (depends on 1)

The method of claim 1, wherein the plurality of data content items comprises a plurality of images, and the characteristic comprises blur.

### Claim 9 (depends on 1)

The method of claim 1, wherein the plurality of data content items comprises a plurality of images, and the characteristic comprises occlusion.

### Claim 10 (depends on 1)

The method of claim 1, wherein the feature is a scale-invariant feature transform codebook feature extracted from a histogram equalization performed on the plurality of media content items.

### Claim 11 (depends on 1)

The method of claim 1, wherein the feature is a scale-invariant feature transform codebook feature extracted from a self-quotient image computed from the plurality of media content items.

### Claim 12 (depends on 1)

The method of claim 1, wherein the feature is a speeded-up robust feature codebook feature extracted from an image deconvolution performed on the plurality of media content items.

### Claim 13 (depends on 1)

The method of claim 1, wherein the feature is a local phase quantized codebook feature extracted from the plurality of media content items.

### Claim 14 (depends on 1)

The method of claim 1, wherein the encoding results a plurality of simulated domains, each domain of the plurality of simulated domains corresponding to a different encoding technique used in the encoding and containing a different feature description corresponding to a feature extracted by the different encoding technique.

### Claim 15 (depends on 14)

The method of claim 14, further comprising:
performing global modeling of the plurality of simulated domains by combining information across the plurality of simulated domains; and
performing local modeling of the plurality of media content items by translating information learned in the global modeling onto the plurality of media content items.

### Claim 16 (depends on 15)

The method of claim 15, wherein the global modeling comprises:
generating a p-dimensional generative sub-space for each domain of the plurality of simulated domains, where each sub-space generated represents a point on a Grassman manifold comprising a space of all p-dimensional sub-spaces.

### Claim 17 (depends on 16)

The method of claim 16, wherein the p-dimensional generative sub-space is generated using principal component analysis.

### Claim 18 (depends on 16)

The method of claim 16, wherein the local modeling comprises:
projecting data contained in each domain of the plurality of simulated domains onto a set of p-dimensional generative sub-spaces corresponding to the plurality of simulated domains to generate a plurality of projections; and
learning a q-dimensional sub-space via grouping of the plurality of projections.

### Claim 19

A device, comprising:
a processor; and
a computer-readable medium storing instructions which, when executed by the processor, cause the processor to perform operations comprising:
acquiring a plurality of data content items from a plurality of different domains, wherein at least some data content items of the plurality of data content items are unlabeled;
encoding the plurality of data content items with a feature representing a domain shift variation that is assumed to be present in the plurality of data content items, wherein the domain shift variation comprises variation in a characteristic of the plurality of data content items; and
clustering the plurality of data content items into a predefined number of content categories subsequent to the encoding.

### Claim 20

A non-transitory computer-readable storage medium storing instructions which, when executed by a processor, cause the processor to perform operations, the operations comprising:
acquiring a plurality of data content items from a plurality of different domains, wherein at least some data content items of the plurality of data content items are unlabeled;
encoding the plurality of data content items with a feature representing a domain shift variation that is assumed to be present in the plurality of data content items, wherein the domain shift variation comprises variation in a characteristic of the plurality of data content items; and
clustering the plurality of data content items into a predefined number of content categories subsequent to the encoding.
