Point-Cloud Connectivity Ordering for Faster Neural Network Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing neural network processing of point clouds is inefficient due to arbitrary data point orders that do not reflect spatial proximity, leading to sparse and inefficient connectivity information, which affects memory locality and processing speed.
Innovation Solution
Determine a new order for data points based on their spatial positions using space-filling curves, generating updated connectivity information that reflects spatial distribution and concentration of connections, improving data structure efficiency for neural network processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If data points are processed in arbitrary order, then processing simplicity is maintained, but memory locality and processing speed deteriorate
Solution Approach 1:
The patent applies preliminary action by determining the optimal order for data points based on their spatial positions before performing neural network processing. This reordering step is executed in advance to optimize memory access patterns and improve processing speed during the actual neural network computation, without complicating the core processing algorithm.
2Productivity
If data points are reordered based on spatial positions, then memory locality and processing speed improve, but computational complexity increases
Solution Approach 1:
The patent changes the ordering parameter of data points from arbitrary indices to spatial position-based indices. By determining the optimal order based on spatial coordinates and using this ordering to reorganize connectivity information, the patent improves memory locality and processing speed while keeping the additional computational overhead limited to the ordering determination step.
3Quantity of substance
If connectivity information is sparse, then data storage efficiency improves, but neural network processing efficiency deteriorates
Solution Approach 1:
The patent merges the ordering information with connectivity information by determining the optimal order for data points and generating updated connectivity information that reflects this ordering. This integration ensures that both data points and their connectivity relationships are organized in the same optimal sequence, improving neural network processing efficiency while maintaining storage efficiency through coordinated reorganization rather than separate optimizations.
Data Source
AI summary
For a set of data points which are desired to be processed according to neural network processing, each data point corresponding to a position in space, data point information indicative of one or more properties of the data points is received (500), and connectivity information indicative of connections between the data points is determined (503). An order for the data points is then determined (504) based on the positions in space of the data points, and updated connectivity information (505) is generated based on the initial connectivity information and the determined order for the set of data points. The updated connectivity information and data point information are provided for further processing (507) to be performed by a processor operable to execute neural network processing.