BUNDLING HYPERVECTORS

Organization Name

INTERNATIONAL BUSINESS MACHINES CORPORATION

Inventor(s)

BUNDLING HYPERVECTORS - A simplified explanation of the abstract

This abstract first appeared for US patent application 17809052 titled 'BUNDLING HYPERVECTORS

Simplified Explanation

The patent application describes a method for bundling a set of code hypervectors using weights and mapping techniques. Here are the key points:

The method involves bundling a set of M code hypervectors, each with a dimension of D, where M is greater than 1.
A weight vector is received, which is an M-dimensional vector containing weights for weighting the set of code hypervectors.
The weight vector is then mapped to an S-dimensional vector, s, where each element of s indicates one of the code hypervectors. The mapping is such that S is equal to D divided by L, and L is greater than or equal to 1.
Finally, a hypervector is built using the mapped vector, where each element of the built hypervector corresponds to the element of the code hypervector indicated in the corresponding element of the S-dimensional vector, s.

Potential applications of this technology:

Pattern recognition: The method can be used to bundle code hypervectors representing patterns or features, allowing for efficient recognition and classification of similar patterns.
Data compression: By bundling multiple code hypervectors into a single hypervector, the method can be used for data compression, reducing storage requirements.
Machine learning: The technique can be applied in machine learning algorithms to improve the efficiency and accuracy of training and inference processes.

Problems solved by this technology:

Efficient representation: The method provides a way to bundle multiple code hypervectors into a single hypervector, reducing the complexity and storage requirements of representing multiple patterns or features.
Weighted bundling: The use of weights allows for prioritization or emphasis on specific code hypervectors, enabling more flexible and customizable bundling.
Mapping technique: The mapping of the weight vector to the S-dimensional vector provides a systematic way to indicate which code hypervectors are included in the bundled hypervector.

Benefits of this technology:

Improved efficiency: The bundling of code hypervectors reduces the computational complexity and storage requirements, leading to faster processing and reduced memory usage.
Customizability: The use of weights and mapping techniques allows for customization and prioritization of code hypervectors, enabling tailored solutions for specific applications.
Versatility: The method can be applied in various domains such as pattern recognition, data compression, and machine learning, making it a versatile solution for different use cases.

Original Abstract Submitted

Embodiments are disclosed for a method. The method includes bundling a set of M code hypervectors, each of dimension D, where M>1. The bundling includes receiving an M-dimensional vector comprising weights for weighting the set of code hypervectors. The bundling further includes mapping the M-dimensional vector to an S-dimensional vector, s, such that each element of the S-dimensional vector, s, indicates one of the set of code hypervectors, where S=D/L and L≥1. Additionally, the bundling includes building a hypervector such that an ith element of the built hypervector is an ith element of the code hypervector indicated in an ith element of the S-dimensional vector, s.