17809044. BLOCKWISE FACTORIZATION OF HYPERVECTORS simplified abstract (INTERNATIONAL BUSINESS MACHINES CORPORATION)
Contents
BLOCKWISE FACTORIZATION OF HYPERVECTORS
Organization Name
INTERNATIONAL BUSINESS MACHINES CORPORATION
Inventor(s)
Michael Andreas Hersche of Zurich (CH)
Abu Sebastian of Adliswil (CH)
Abbas Rahimi of Rüschlikon (CH)
BLOCKWISE FACTORIZATION OF HYPERVECTORS - A simplified explanation of the abstract
This abstract first appeared for US patent application 17809044 titled 'BLOCKWISE FACTORIZATION OF HYPERVECTORS
Simplified Explanation
The abstract describes a method for determining the cognitive concepts represented by a data structure using hypervectors. Here are the key points:
- The method determines the granularity of hypervectors, which are mathematical representations of data structures.
- It receives an input hypervector that represents a data structure.
- The method performs an iterative process to break down the input hypervector into individual hypervectors that represent cognitive concepts.
- For each concept, an unbound version of a hypervector representing the concept is determined by comparing it with estimate hypervectors of other concepts.
- A similarity vector is generated to indicate the similarity between the unbound version of the hypervector and each candidate code hypervector of the concept.
- An estimate of a hypervector representing the concept is generated by combining the candidate code hypervectors and the weights of the similarity vector.
Potential applications of this technology:
- Cognitive computing: This method can be used to analyze and understand complex data structures, such as natural language processing or image recognition.
- Artificial intelligence: The method can help in building intelligent systems that can learn and recognize patterns in data.
- Data analysis: By breaking down data structures into cognitive concepts, this method can assist in analyzing and extracting meaningful insights from large datasets.
Problems solved by this technology:
- Complex data analysis: The method provides a systematic approach to break down and analyze complex data structures, making it easier to understand and extract useful information.
- Pattern recognition: By representing data structures as hypervectors and breaking them down into cognitive concepts, the method enables the recognition of patterns and similarities in the data.
Benefits of this technology:
- Improved data understanding: The method helps in understanding the underlying cognitive concepts represented by data structures, leading to better insights and decision-making.
- Efficient data analysis: By factorizing the input hypervector into individual hypervectors, the method simplifies the analysis process and reduces computational complexity.
- Scalability: The method can be applied to large datasets and can handle a wide range of data structures, making it scalable for various applications.
Original Abstract Submitted
Embodiments are disclosed for a method. The method includes determining a granularity of hypervectors. The method also includes receiving an input hypervector representing a data structure. Additionally, the method includes performing an iterative process to factorize the input hypervector into individual hypervectors representing the cognitive concepts. The iterative process includes, for each concept: determining an unbound version of a hypervector representing the concept by a blockwise unbinding operation between the input hypervector and estimate hypervectors of other concepts. The iterative process further includes determining a similarity vector indicating a similarity of the unbound version of the hypervector with each candidate code hypervector of the concept. Additionally, the iterative process includes generating an estimate of a hypervector representing the concept by a linear combination of the candidate code hypervectors, and weights of the similarity vector.
