1

Data Generation

@Sridhar Rao

Failure and Anomaly Data Generation 

The Data can also be used for ITU challenge.

Please share your inputs in the above link - Its an experiment!

2

Log-Analysis - Status Update

@Rohit Singh Rathaur

Moved to next week.

3

Update on KubeFlow and GANs

@Girish

Trying: Acumos and Kubeflow.

Kubeflow: Ease to use. Support multiple cloud (AWS, GCP).  First the K8S cluster is deployed, and on that Kubeflow Pods are deployed. arrikto - Knowledge sharing on Kubeflow. By default, simple models (linear regression, SVMs, etc) are available. Library support: TF and PyTorch. 

Data-Pipeline Integration Support: Prometheus. Elasticsearch.

ToDo: Migrate the FP model to KubeFlow.

Next Week: How to Add new Model to Kubeflow.?

GANs: Slight improvement with Cycle-GANs, work is ongoing.

4

ITU Challenge Update

@Sridhar Rao

Document to submit ready. We will ready with the data by March.

5