Topics

EdgeX Demo & ML

Tiejun Chen
 

James and Brett,

Talking about center stage demo, I'm thinking if we can bring EdgeX Foundry and ML together. EdgeX itself doesn't have sort of intelligence at edge side unless we use on-gateway client to do decision-making. Actually, there are a lot of Java ML libraries so it's not difficult to combine EdgeX and ML. It could be a compelling use case.

Furthermore, I also believe we probably can consider that EdgeX should provide a common interface to support ML. It can help everyone to integrate/develop his own ML library based on EdgeX. It might be a plugin framework. If you guys this can be a component of EdgeX, I'd like to follow this.

Thanks
Tiejun

Philip DesAutels <pdesautels@...>
 

I think you are hitting on a key area that we need to bring into the broader community discussion of EdgeX - specifically what is the ML story and architecture for EdgeX - I believe that the discussion best land in Applications Working Group: https://wiki.edgexfoundry.org/display/FA/Applications+Working+Group

On Fri, Aug 18, 2017 at 10:02 AM, Tiejun Chen <tiejunc@...> wrote:
James and Brett,

Talking about center stage demo, I'm thinking if we can bring EdgeX Foundry and ML together. EdgeX itself doesn't have sort of intelligence at edge side unless we use on-gateway client to do decision-making. Actually, there are a lot of Java ML libraries so it's not difficult to combine EdgeX and ML. It could be a compelling use case.

Furthermore, I also believe we probably can consider that EdgeX should provide a common interface to support ML. It can help everyone to integrate/develop his own ML library based on EdgeX. It might be a plugin framework. If you guys this can be a component of EdgeX, I'd like to follow this.

Thanks
Tiejun



--
Philip

Philip DesAutels, PhD
Senior Director of IoT
The Linux Foundation
+1.781.363.3713   skype:philip.desautels