Sedna joint inference and federated learning controller optimization

What would you like to be added/modified: Sedna is an edge-cloud synergy AI project incubated in KubeEdge SIG AI. Benefiting from the edge-cloud synergy capabilities provided by KubeEdge, Sedna can implement across edge-cloud collaborative training and collaborative inference capabilities, such as joint inference, incremental learning, federated learning, and lifelong learning. Sedna supports popular AI frameworks, such as TensorFlow, Pytorch, PaddlePaddle, MindSpore.

Sedna can simply enable edge-cloud synergy capabilities to existing training and inference scripts, bringing the benefits of reducing costs, improving model performance, and protecting data privacy. However, there are still some functional defects in the joint inference and federated learning controller in the current Sedna project, which need to be solved, mainly in the following aspects: Joint inference: 1. after the creation of joint inference task or federated learning task, the generated cloud and edge task instances will not be automatically rebuilt after failure or manual deletion, that is, lack of self-healing ability; 2. After the joint inference task CR is deleted, the task instance and service configuration generated by CR will not be cascaded. This defect will cause the subsequent failure to create the joint inference task. What needs to be done: Current Sedna's joint inference and federated learning controllers are optimized to address the above functional deficiencies.

Why is this needed: Current bugs in joint inference and federated learning can seriously affect the normal operation of both.

Recommended Skills: Golang / Python

Useful links: https://github.com/kubernetes/kubernetes https://kubernetes.io/ https://github.com/kubernetes/client-go https://github.com/kubeedge/kubeedge https://github.com/kubeedge/sedna https://www.topgoer.com/

kubeedge / sedna

Sedna joint inference and federated learning controller optimization #430