20240022787. CARRIAGE AND SIGNALING OF NEURAL NETWORK REPRESENTATIONS simplified abstract (Nokia Technologies Oy)

From WikiPatents
Jump to navigation Jump to search

CARRIAGE AND SIGNALING OF NEURAL NETWORK REPRESENTATIONS

Organization Name

Nokia Technologies Oy

Inventor(s)

Emre Aksu of Tampere (FI)

Miska Hannuksela of Tampere (AD)

[[:Category:Francesco Cricr� of Tampere (FI)|Francesco Cricr� of Tampere (FI)]][[Category:Francesco Cricr� of Tampere (FI)]]

Hamed Rezazadegan Tavakoli of Espoo (FI)

CARRIAGE AND SIGNALING OF NEURAL NETWORK REPRESENTATIONS - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240022787 titled 'CARRIAGE AND SIGNALING OF NEURAL NETWORK REPRESENTATIONS

Simplified Explanation

A method is provided for defining a metadata box of a neural network representation (nnr) item data, where the nnr item data includes an nnr bitstream. The method involves defining an association between the nnr item data and an nnr configuration using a configuration item property, which contains information about the stored nnr item data. Corresponding apparatuses and computer program products are also provided.

  • The method defines a metadata box for nnr item data in a neural network representation.
  • The nnr item data includes an nnr bitstream.
  • An association is established between the nnr item data and an nnr configuration.
  • The association is defined using a configuration item property.
  • The configuration item property contains information about the stored nnr item data.
  • The method can be implemented using corresponding apparatuses and computer program products.

Potential applications of this technology:

  • Neural network representation in various fields such as computer vision, natural language processing, and speech recognition.
  • Efficient storage and retrieval of nnr item data.
  • Integration of nnr item data with nnr configurations for better management and analysis.

Problems solved by this technology:

  • Lack of standardized methods for defining metadata in neural network representations.
  • Difficulty in associating nnr item data with nnr configurations.
  • Inefficient storage and retrieval of nnr item data.

Benefits of this technology:

  • Improved organization and management of nnr item data.
  • Enhanced analysis and understanding of neural network representations.
  • Efficient storage and retrieval of nnr item data.
  • Standardized approach for defining metadata in neural network representations.


Original Abstract Submitted

a method is provided for defining a metadata box of a neural network representation (nnr) item data, wherein the nnr item data comprises an nnr bitstream; and defining an association between the nnr item data and an nnr configuration by using a configuration item property, wherein the nnr configuration item property comprises information about stored nnr item data. corresponding apparatuses and computer program products are also provided.