FRAMEWORK FOR MANAGING MACHINE LEARNER PERCEPTION TASKS ON LIMITED HARDWARE RESOURCES

Abstract

Techniques for managing machine learning (ML) perception tasks on limited hardware resources are revealed here.An example system includes one or more processors that execute instructions to: register a specification of semantic output to be produced by the ML task processor in a register for each of a plurality of ML task processors; receive a request for a specific semantic output from a task manager module and a client; identify an ML task processor based on the register, where the specification of semantic output for the ML task processor matches the specified semantic output; in response to determining that the ML task processor is currently running, refrain from loading the ML task processor; in response to determining that the ML task processor is currently not running, execute the ML task processor; and establish a connection between the client and the ML task processor to allow the client to receive the semantic output.

Images