18326800. EFFICIENT MACHINE LEARNING MESSAGE PASSING ON POINT CLOUD DATA simplified abstract (QUALCOMM Incorporated)
Contents
EFFICIENT MACHINE LEARNING MESSAGE PASSING ON POINT CLOUD DATA
Organization Name
Inventor(s)
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 18326800 titled 'EFFICIENT MACHINE LEARNING MESSAGE PASSING ON POINT CLOUD DATA
Simplified Explanation
The present disclosure relates to techniques and apparatus for improving machine learning. The patent application describes a method for processing input data in multidimensional space using a neural network.
- Input data consisting of multiple points in 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 processed by 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.
- An output inference is generated based on the output feature.
Potential applications of this technology:
- Machine learning and data analysis: The techniques described in the patent application can be applied to various machine learning tasks, such as image recognition, natural language processing, and anomaly detection.
- Pattern recognition: The method can be used to identify patterns and relationships in complex datasets, enabling more accurate predictions and classifications.
- Data visualization: By mapping edges to defined axes in multidimensional space, the technology can help visualize and understand high-dimensional data.
Problems solved by this technology:
- Improved feature extraction: The method allows for the generation of intermediate features that capture important information from the input data, leading to more accurate and meaningful representations.
- Dimensionality reduction: By mapping edges to defined axes, the technique helps reduce the dimensionality of the data, making it easier to analyze and interpret.
- Robustness to transformations: The use of group elements and their inverses ensures that the method is invariant to certain transformations, making it more robust and applicable to a wide range of datasets.
Benefits of this technology:
- Enhanced machine learning performance: The improved feature extraction and dimensionality reduction capabilities of the method can lead to better machine learning models with higher accuracy and efficiency.
- Increased interpretability: By mapping edges to defined axes, the technique provides a more interpretable representation of the data, allowing for better understanding and insights.
- Robustness and versatility: The use of group elements and their inverses ensures that the method can handle various types of data and is invariant to certain transformations, making it applicable to diverse real-world scenarios.
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.
