Deep learning architecture for edge computing system
AssigneeAmazon Technologies, Inc.
InventorsVivek Viswanathan, Senthil Natarajan
Life of this patent
Estimating the term…
Abstract
Devices and techniques are generally described for an edge computing system. In various examples, a first application executed by a first virtual machine may generate a configuration file specifying a first machine learning task and first data. The first virtual machine may send the configuration file to a second virtual machine. The second virtual machine may generate first data effective to cause the first machine learning task to be executed by a third virtual machine. In various examples, the second virtual machine may retrieve the first data. In at least some examples, the second virtual machine may send the first data and the second data to a third virtual machine configured with access to a graphical processing unit.
Claims20
A computing device comprising:
- at least one processor;
- a graphical processing unit;
- and
- at least one memory configured in communication with the at least one processor, the at least one memory storing instructions that, when executed by the at least one processor, are effective to program the at least one processor to:
- execute a first virtual machine, the first virtual machine effective to:
- execute a first application;
- generate, by the first application, a configuration file specifying a first computer vision operation;
- and
- send the configuration file to a second virtual machine, wherein the second virtual machine is configured to lack Internet access;
- wherein the second virtual machine is effective to:
- determine, from the configuration file, first data associated with the first computer vision operation;
- determine, from the configuration file, at least one frame of first video data on which to perform the first computer vision operation;
- generate decoded video data by decoding the at least one frame of first video data;
- store the decoded video data in the at least one memory;
- receive, from the at least one memory, the decoded video data;
- generate an inference request comprising a first request to perform the first computer vision operation on the decoded video data;
- and
- send the inference request to a third virtual machine, wherein the third virtual machine is configured with access to the graphical processing unit.
A method comprising:
- generating, using a first application executed by a first virtual machine executing on a computing device, (i) a configuration file specifying a first machine learning task and (ii) first data;
- sending, by the first virtual machine, the configuration file to a second virtual machine executing on the computing device, wherein the second virtual machine is configured to lack Internet access;
- generating, using the second virtual machine, second data effective to cause the first machine learning task to be executed by a third virtual machine executing on the computing device using at least one of a graphical processing unit (GPU) or an accelerator circuit accessible by the third virtual machine;
- receiving, by the second virtual machine, the first data;
- and
- sending, by the second virtual machine, the first data and the second data to the third virtual machine.
A computing device, comprising:
- at least one processor;
- and
- at least one memory configured in communication with the at least one processor, the at least one memory storing instructions that, when executed by the at least one processor, are effective to program the at least one processor to:
- generate, using a first application executed by a first virtual machine, a configuration file specifying a first machine learning task and first data;
- send, by the first virtual machine, the configuration file to a second virtual machine, wherein the second virtual machine is configured to lack Internet access;
- generate, using the second virtual machine, second data effective to cause the first machine learning task to be executed by a third virtual machine using at least one of a graphical processing unit (GPU) or an accelerator circuit accessible by the third virtual machine;
- receive, by the second virtual machine, the first data;
- and
- send, by the second virtual machine, the first data and the second data to the third virtual machine.
Dependencies are inferred from the claim text.