Translating sensory communications

Patent metadata

Field	Value
Publication number	US-12542835-B2
Application number	US-202118292118-A
Country	US
Kind code	B2
Filing date	Jul 27, 2021
Priority date	Jul 27, 2021
Publication date	Feb 3, 2026
Grant date	Feb 3, 2026

Abstract

A computer implemented method for translating a sensory input to a sensory output for communication between first and second entities. The first entity includes a user and the second entity includes a user or a computing system. The method includes obtaining a first profile specifying a communication capability of the first entity, and a second profile specifying a communication capability of the second entity and obtaining a trained Machine Learning (ML) model operable to map an input sensory communication in accordance with one of the first or second profiles to an output sensory communication in accordance with the other of the profiles. The method further includes receiving an input sensory communication from one of the first or second entities, using the ML model to map the input sensory communication to an output sensory communication, and providing the output sensory communication to the other of the entities.

First claim

The invention claimed is:

1 . A computer implemented method for translating a sensory input to a sensory output for communication between a first entity and a second entity, the first entity comprising a first user and the second entity comprises a second user or a computing system, the method, performed by a translation system, comprising:
obtaining a first profile specifying a communication capability of the first entity, and a second profile specifying a communication capability of the second entity; obtaining a trained Machine Learning, ML, model configured to map an input sensory communication in accordance with one of the first profile or second profile to an output sensory communication in accordance with the other of the first profile or second profile; receiving an input sensory communication from one of the first entity or second entity; using the ML model to map the input sensory communication to an output sensory communication; and providing the output sensory communication to the other of the first entity or second entity; wherein obtaining a trained ML model configured to map an input sensory communication in accordance with one of the first profile or second profiles to an output sensory communication in accordance with the other of the first profile or second profile comprises: obtaining, from a repository, a candidate set of at least one trained ML model configured to map an input sensory communication in accordance with one of the first profile or second profile to an output sensory communication in accordance with the other of the first profile or second profile; and selecting, from the candidate set, an ML model for use in translation of communication between the first and second entities.

2 . The method as claimed in claim 1 , wherein the input sensory communication is received from an entity comprising a user; and wherein receiving the input sensory communication comprises receiving a representation, generated by a user device of the user, of a sensory communication input by the user to the user device.

3 . The method as claimed in claim 1 , wherein the other of the first entity or second entity comprises a user; and wherein providing the output sensory communication to the other of the first entity or second entity comprises providing to a user device of the user a representation of the output sensory communication, wherein the user device is configured to generate the output sensory communication for the user from the provided representation.

4 . The method as claimed in claim 1 , wherein a communication capability comprises at least one of a sensory capability or an actuation capability.

5 . The method as claimed in claim 1 , wherein a sensory communication comprises a communication the reception and interpretation of which requires a sensory capability.

6 . The method as claimed in claim 5 , wherein the received input sensory communication and the provided output sensory communication require different sensory capabilities for reception and interpretation.

7 . The method as claimed in claim 1 , wherein a sensory communication in accordance with a profile of an entity comprises at least one of: a sensory communication in a format that is compatible with a sensory capability of the entity; or a sensory communication in a format from which or to which the entity, or a user device associated with the entity, is configured to translate.

8 . The method as claimed in claim 1 , wherein obtaining at least one of a first profile or second profile specifying a communication capability of the first or second entity comprises at least one of:
retrieving the first or second profile from a memory; receiving the first or second profile in a message; and detecting component elements of the first profile or second profile using a sensor.

9 . The method as claimed in claim 1 , wherein selecting, from the candidate set, an ML model for use in translation of communication between the first entity and second entity comprises selecting an ML model using feedback provided for ML models in the candidate set from users of the ML models.

10 . The method as claimed in claim 1 , further comprising obtaining a context for the communication between first entity and second entity; and wherein selecting, from the candidate set, an ML model for use in translation of communication between first entity and second entity comprises using the obtained context to select the ML model.

11 . The method as claimed in claim 10 , wherein the obtained context comprises an identification of a task to be achieved through the communication between the first entity and second entity.

12 . The method as claimed in claim 1 , further comprising:
receiving an authentication token from at least one of the first entity and second entity; and using the authentication token to authenticate the first or second entity.

13 . The method as claimed in claim 1 , wherein using the ML model to map the input sensory communication to an output sensory communication comprises: inputting to the ML model a representation of the input sensory communication, wherein the representation is processed by the ML model in accordance with trained values of the ML model parameters; and obtaining from the ML model an output sensory communication comprising a representation of information contained in the input sensory communication.

14 . The method as claimed in claim 13 , wherein the representation of information contained in the input sensory communication comprises at least one of: a representation of the information in a format that is compatible with a sensory capability of the entity to which the output sensory communication is provided; or a sensory communication in a format from which the entity to which the output sensory communication is provided, or a user device associated with the entity, is configured to translate.

15 . The method as claimed in claim 1 , wherein the ML model comprises at least one of:
a multi-input-multi output Neural Network; a split Neural Network; a conditional Variational AutoEncoder, cVAE; a generative model; and a Natural Language Processing model.

16 . The method as claimed in claim 1 , further comprising:
receiving feedback from at least one of the first or second entities; and performing at least one of: providing the feedback to a training node for inclusion in a training data set for the obtained ML model; providing the feedback to a management node for inclusion in a training data set for a selection model for selecting ML models for translating a sensory input to a sensory output for communication between first entity and second entity; using the feedback in a training process to update the obtained ML model; or using the feedback in a training process to update a model for selecting ML models for translating a sensory input to a sensory output for communication between first entity and second entity.

17 . The method as claimed in claim 1 , wherein the input sensory communication requires a sensory capability for reception and interpretation that is:
not present in the profile of the entity to which the output sensory communication is provided; identified in the profile of the entity to which the output sensory communication is provided as non-optimal under at least some conditions; or
identified in the profile of the entity to which the output sensory communication is provided as unavailable under at least some conditions.

18 . The method as claimed in claim 1 , wherein the output sensory communication requir

Assignees

• Ericsson Telefon Ab L M

Inventors

• Ickin Selim
• Azari Amin
• Lee Altmann Carmen
• Mujumdar Anusha Pradeep
• Taghia Jalil

Classifications

• G06N3/0475

  Generative networks · CPC title

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title

• H04L67/12

  specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks · CPC title

• H04L67/565 Primary

  Conversion or adaptation of application format or content (adding application control or application functional data H04L67/561) · CPC title

• G06N3/09

  Supervised learning · CPC title