It would be good to have a few more example models. One option is to use the Pipelines exercise from AMPCamp 5 to do image classification, and then add a personalization component for the final step of the prediction. This uses the Cifar-10 dataset, which has the advantage that the pictures are small and easy to display visually. This model should be added to the veloxms-examples module.
I don't think we're going to get to this so I'm removing it from the Alpha Release milestone. Having an integration with the Keystone-ML project is even better though.
It would be good to have a few more example models. One option is to use the Pipelines exercise from AMPCamp 5 to do image classification, and then add a personalization component for the final step of the prediction. This uses the Cifar-10 dataset, which has the advantage that the pictures are small and easy to display visually. This model should be added to the
veloxms-examples
module.