Sign in Subscribe
Software industry

Cloud Instance Sizing and Deployments Using Machine Learning

Strategic Insights

— 2 min read

The inventors Dinh, Hung, Linsey, David J., and Mohanty, Bijan Kumar have developed a cloud instance prediction platform that utilizes machine learning algorithms to predict the configuration of a cloud instance based on the features of an application. The platform aims to simplify the process of selecting the virtual environments in which applications may run, given the complexity of virtual environment selection in cloud computing.

Key Takeaways:

  • The platform receives a request to predict a configuration of a cloud instance in which at least one application is to be executed, and analyzes the features of the application using one or more machine learning algorithms.
  • The machine learning algorithms comprise a neural network that predicts a plurality of targets, with the first target being predicted using a classification technique and the remaining targets being predicted using a regression technique.
  • The platform is configured to interface with at least one cloud platform to collect one or more runtime metrics corresponding to execution of a plurality of applications in a plurality of cloud instances.
  • The neural network includes a plurality of parallel branches corresponding to the plurality of targets, with respective branches comprising at least two hidden layers utilizing a rectified linear unit activation function.
  • The platform can be trained with historical runtime feature data of a plurality of applications.
  • The method comprises selecting a cloud platform of a plurality of cloud platforms to host the cloud instance based on the analyzing.
  • The apparatus comprises a processing device operatively coupled to a memory and configured to receive a request to predict a configuration of a cloud instance in which at least one application is to be executed.
  • The article of manufacture comprises a non-transitory processor-readable storage medium having stored therein program code that causes a processing device to perform the steps of the method.

Statistics:

  • The platform predicts the configuration of a cloud instance in which at least one application is to be executed based on the analyzing of one or more features using one or more machine learning algorithms.
  • The machine learning algorithms comprise a neural network that predicts a plurality of targets, with the first target being predicted using a classification technique and the remaining targets being predicted using a regression technique.
  • The platform is configured to interface with at least one cloud platform to collect one or more runtime metrics corresponding to execution of a plurality of applications in a plurality of cloud instances.
  • The neural network includes a plurality of parallel branches corresponding to the plurality of targets, with respective branches comprising at least two hidden layers utilizing a rectified linear unit activation function.

Sources:

  • Dinh, Hung; Linsey, David J.; Mohanty, Bijan Kumar. Cloud Instance Sizing And Deployments Using Machine Learning. U.S. Patent Application Number 20250310194, filed March 27, 2024 and posted October 2, 2025. Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(20250310194)&db=US-PGPUB&type=ids
  • NewsRx LLC. 2025. Cloud Instance Sizing and Deployments Using Machine Learning. VerticalNews.

Read more