20240037453. EFFICIENT MACHINE LEARNING MESSAGE PASSING ON POINT CLOUD DATA simplified abstract (QUALCOMM Incorporated)

EFFICIENT MACHINE LEARNING MESSAGE PASSING ON POINT CLOUD DATA

Organization Name

QUALCOMM Incorporated

Inventor(s)

Pim De Haan of Amsterdam (NL)

Taco Sebastiaan Cohen of Amsterdam (NL)

EFFICIENT MACHINE LEARNING MESSAGE PASSING ON POINT CLOUD DATA - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240037453 titled 'EFFICIENT MACHINE LEARNING MESSAGE PASSING ON POINT CLOUD DATA

Simplified Explanation

The present disclosure provides techniques and apparatus for improved machine learning. The abstract describes a method where input data consisting of multiple points in a multidimensional space is accessed. An edge connecting two points is identified and mapped to a defined axis in the multidimensional space using a group element. The mapped edge is then processed using a neural network to generate an intermediate feature. An output feature is generated by applying the inverse of the group element to the intermediate feature. Finally, an output inference is generated based on the output feature.

• Techniques and apparatus for improved machine learning
• Accessing input data consisting of multiple points in a multidimensional space
• Identifying and mapping an edge connecting two points to a defined axis using a group element
• Processing the mapped edge using a neural network to generate an intermediate feature
• Generating an output feature by applying the inverse of the group element to the intermediate feature
• Generating an output inference based on the output feature

Potential applications of this technology:

• Enhancing machine learning algorithms and models
• Improving pattern recognition and classification tasks
• Enhancing data analysis and decision-making processes
• Optimizing neural network architectures and training methods

Problems solved by this technology:

• Overcoming limitations in traditional machine learning approaches
• Addressing challenges in processing multidimensional data
• Improving the accuracy and efficiency of machine learning algorithms
• Enhancing the interpretability and generalization capabilities of models

Benefits of this technology:

• Improved accuracy and performance of machine learning models
• Enhanced understanding and interpretability of complex data patterns
• Increased efficiency and speed in processing multidimensional data
• Potential for discovering new insights and knowledge from large datasets

Original Abstract Submitted

certain aspects of the present disclosure provide techniques and apparatus for improved machine learning. input data comprising a plurality of points in multidimensional space is accessed. an edge connecting a first point and a second point of the plurality of points is identified, and the edge is mapped to a defined axis in the multidimensional space by applying a group element to the edge. an intermediate feature is generated by processing the mapped edge using a neural network. an output feature is generated by applying an inverse of the group element to the intermediate feature, and an output inference is generated based at least in part on the output feature.