Back to results
US 12,755,141 B2Granted
Mesh based convolutional neural network techniques
AssigneeNVIDIA Corporation
InventorsNuttapong Chentanez, Miles Macklin, Matthias Mueller-Fischer, Stefan Jeschke, Tae-Yong Kim
Life of this patent
Estimating the term…
Abstract
Convolutional operators for triangle meshes are determined to construct one or more neural networks. In at least one embodiment, convolutional operators, pooling operators, and unpooling operators are determined to construct the one or more neural networks, in which the same learned weights from the one or more neural networks can further be used for triangle meshes with different topologies.
Claims29
A system, comprising:
- one or more computers having one or more processors to train a neural network by:
- performing a convolution on data input, at a layer of the neural network, wherein the convolution is performed by using convolutional operators on the data input, the convolutional operators having been determined by:
- selecting a vertex from a plurality of vertices of the data input;
- obtaining a plurality of sampling points positioned uniformly around the selected vertex based on a length of the convolutional operators, at least a portion of the plurality of sampling points being other than one of the plurality of vertices;
- adding at least a portion of the plurality of vertices to a list such that each vertex in the portion of the plurality of vertices is or is paired with a sampled point from the plurality of sampling points;
- and
- determining, based at least in part of the list, a set of vertices to form the convolutional operators;
- and
- using the convolutional operators on the data input to generate a set of outputs of the convolution.
6 more elements
A processor comprising:
- one or more arithmetic logic units (ALUs) to train one or more neural networks, at least in part, by:
- determining one or more convolutional operators to perform a convolution on received data input, wherein the determination of the one or more convolutional operators was performed by:
- obtaining a plurality of sampling points positioned uniformly around for a vertex of a plurality of vertices of the data input, at least a portion of the plurality of sampling points being other than one of the plurality of vertices;
- generating an index to indicate that the vertex and at least a portion of the plurality of vertices is or is paired with a sampled point from the plurality of sampling points;
- and
- determining the one or more convolutional operators using information from the index.
3 more elements
A method, comprising:
- training one or more neural networks by determining convolutional operators to perform a convolution on data, wherein the convolutional operators are determined by:
- obtaining a plurality of sampling points positioned uniformly around for a vertex of a plurality of vertices of the data, based on a length of the convolutional operators, at least a portion of the plurality of sampling points being other than one of the plurality of vertices;
- adding at least a portion of the plurality of vertices to a list such that each vertex in the portion of the plurality of vertices is or is paired with a sampled point from the plurality of sampling points;
- and
- determining, based at least in part of the list, a set of vertices to form the convolutional operators based at least in part on the plurality of vertices and the plurality of sampling points.
2 more elements
Dependencies are inferred from the claim text.