Federated Learning with Edge Impulse

#1

Is it possible within the Edge Impulse Framework to combine privacy and latency compartmentalized models on device with aggregated models in the cloud,

#2

Hello @IML2 ,

Which aspect of Federated Learning are you interested in?
We’d love to hear more about your use case. However, at the moment we don’t do on the device training. All the training is done in the cloud and then your deployed model can run on the edge device without any cloud or internet interaction.

Regards,

Louis

#3

@louis Hello, is federated learning on the roadmap of Edge Impulse?

#4

In meantime I found a nice project on EI: A Federated Approach to Train and Deploy Machine Learning Models

#5

@louis If one deployed an Impulse to the Brainchip AKD1000, then would one be able to execute the Brainchip on-chip learning feature?

Or does one need to build a Model with some other Studio like tool to leverage the Brainchip on-chip learning feature?

#6

and
A Federated Approach to Train and Deploy Machine Learning Models (Flower Monthly 2023-10)