Dynamic machine learning on premise model selection based on entity clustering and feedback

A non-transitory computer-readable medium comprising instructions for providing machine learned models to an on premise computing device, the instructions, when executed by a computing system, cause the computing system to perform operations comprising:

• clustering other customer data including similar telemetric parameters, wherein the customer data is stored on a computing network, the similar telemetric parameters including client count, detected interference, detected power, throughput, number of clients performing onboarding, and average onboarding time;
• generate, based on the clustered other customer data, at least first and second machine learned models, the first machine learned model requiring a predefined minimum of a first amount of computational resources needed to execute the first machine learned model, the second machine learned model requiring a predefined minimum of a second amount of computational resources needed to execute the second machine learned model, the first amount being greater than the second amount;
• and
• provide the first and second machine learned models to an on premise computing device, wherein the on premise computing device:
• selects the first machine learned model in response to at least the available computational resources of the on premise computing device being greater than the first amount of computational resources;
• receives negative feedback, from a customer, wherein the feedback states the customer's current user experience of the on premise device with the first machine learned model is unsatisfactory;
• and
• selects the second machine learned model in response to at least the received negative feedback and the available computational resources of the on premise computing device being greater than the second amount of computational resources.

A system for providing machine learned models to an on premise computing device, the system comprising:

• a processor;
• and
• a non-transitory computer-readable medium storing instructions that, when executed by the system, cause the system to:
• cluster other customer data including similar telemetric parameters, wherein the customer data is stored on a computing network, the similar telemetric parameters including client count, detected interference, detected power, throughput, number of clients performing onboarding, and average onboarding time;
• generate, based on the clustered other customer data, at least first and second machine learned models for the cluster of the other customer data, the first machine learned model requiring a predefined minimum of a first amount of computational resources needed to execute the first machine learned model, the second machine learned model requiring a predefined minimum of a second amount of computational resources needed to execute the second machine learned model, the first amount being greater than the second amount;
• and
• provide the first and second machine learned models to an on premise computing device, wherein the on premise computing device:
• selects the first machine learned model in response to at least the available computational resources of the on premise computing device being greater than the first amount of computational resources;
• receives negative feedback, from a customer, wherein the feedback states the customer's current user experience of the on premise device with the first machine learned model is unsatisfactory;
• and
• selects the second machine learned model in response to at least the received negative feedback and the available computational resources of the on premise computing device being greater than the second amount of computational resources.