18462178. INFERENCE WITH INLINE REAL-TIME ML MODELS IN APPLICATIONS simplified abstract (Microsoft Technology Licensing, LLC)

From WikiPatents
Jump to navigation Jump to search

INFERENCE WITH INLINE REAL-TIME ML MODELS IN APPLICATIONS

Organization Name

Microsoft Technology Licensing, LLC

Inventor(s)

Xenofon Foukas of Cambridge (GB)

Bozidar Radunovic of Cambridge (GB)

INFERENCE WITH INLINE REAL-TIME ML MODELS IN APPLICATIONS - A simplified explanation of the abstract

This abstract first appeared for US patent application 18462178 titled 'INFERENCE WITH INLINE REAL-TIME ML MODELS IN APPLICATIONS

Simplified Explanation

The patent application describes a method for using codelets within applications to apply machine-learning models to application data.

  • Codelets are small pieces of code that can be dynamically loaded into applications during execution.
  • A controller verifies the safety of the codelet based on its bytecode.
  • The verified codelet is loaded into the application's library and executed to apply application data to a machine-learning model.
  • The machine-learning model can be implemented within the codelet or called from a supported machine-learning model in the application's controller.
  • The computing device can reconstruct the machine-learning model based on a serial representation.

Potential Applications

  • This technology can be applied in various industries such as healthcare, finance, and manufacturing where machine-learning models are used to analyze application data.
  • It can be used in real-time data analysis applications to quickly infer results based on the latest data.

Problems Solved

  • The method ensures that codelets used for machine-learning inference within applications meet safety requirements, preventing potential security risks.
  • It allows for dynamic loading of codelets, enabling flexibility and adaptability in applying machine-learning models to different application scenarios.

Benefits

  • The use of codelets allows for efficient execution of machine-learning models within applications.
  • The verification process ensures the safety and reliability of the codelets used.
  • The ability to reconstruct machine-learning models based on a serial representation provides flexibility in model deployment and management.


Original Abstract Submitted

Described are examples for using codelets executing within applications to use machine-learning (ML) models to infer a result based on application data. The codelets may be dynamically loaded into the applications during execution. A controller verifies, based on extended Berkeley packet filter (eBPF) bytecode of the codelet, that the codelet satisfies safety requirements for execution within the application. A computing device executing the application loads the verified codelet into a library of the application. The application executes the verified codelet to apply application data to the machine-learning model to infer a result. The ML model may be implemented by the eBPF code of the codelet or the codelet may include a call to a machine-learning model of a type supported by a controller of the application and a map for a serial representation of the machine-learning model. The computing device may reconstruct the ML model based on the serial representation.

Retrieved from "http://wikipatents.org/index.php?title=18462178._INFERENCE_WITH_INLINE_REAL-TIME_ML_MODELS_IN_APPLICATIONS_simplified_abstract_(Microsoft_Technology_Licensing,_LLC)&oldid=17133"