US 11,886,983 B2 · Reducing hardware resource utilization for residual neural networks

Abstract

Embodiments of the present disclosure include systems and methods for reducing hardware resource utilization by residual neural networks. In some embodiments, a first matrix is received at a layer included in a neural network. The first matrix is compressed to produce a second matrix. The second matrix has a reduced dimensionality relative to a dimensionality of the first matrix. The second matrix is processed through a network block in the layer included in the neural network. The processed second matrix is expanded to produce a third matrix. The third matrix has a dimensionality that is equal to a dimensionality of the first matrix. The third matrix is added to the first matrix to produce a fourth matrix.

Claims20

Dependent claims

A system comprising:

a set of processing units;
and
a non-transitory machine-readable medium storing instructions that when executed by at least one processing unit in the set of processing units cause the at least one processing unit to:
receive, at a layer included in a neural network, a first matrix;
compress the first matrix to produce a second matrix to reduce an amount of hardware resources utilized to process the second matrix, the second matrix having a reduced dimensionality relative to a dimensionality of the first matrix;
process the second matrix through a network block in the layer included in the neural network;
expand the processed second matrix to produce a third matrix, the third matrix having a dimensionality that is equal to a dimensionality of the first matrix;
and
add the third matrix to the first matrix to produce a fourth matrix.

6 more elements

A non-transitory machine-readable medium storing a program executable by at least one processing unit of a computer system, the program comprising sets of instructions for:

receiving, at a layer included in a neural network, a first matrix;
compressing the first matrix to produce a second matrix to reduce an amount of hardware resources utilized to process the second matrix, the second matrix having a reduced dimensionality relative to a dimensionality of the first matrix;
processing the second matrix through a network block in the layer included in the neural network;
expanding the processed second matrix to produce a third matrix, the third matrix having a dimensionality that is equal to a dimensionality of the first matrix;
and
adding the third matrix to the first matrix to produce a fourth matrix.

3 more elements

Dependencies are inferred from the claim text.

Reducing hardware resource utilization for residual neural networks

Life of this patent

Reducing hardware resource utilization for residual neural networks

Life of this patent

Abstract

Claims20