We are thinking about building a framework that can be used for unlabeled data selection (using active learning) to reduce human labeling budgets. And hope it can be easily used as a service, currently we are adding more SOTA active learning algorithms and there is a demo for you to play with!
curl \ -X POST http://13.213.8.21:8081/post \ -H 'Content-Type: application/json' \ -d '{"data":[{"uri": "https://www.cs.toronto.edu/~kriz/cifar-10-sample/airplane1.p..."}, {"uri": "https://www.cs.toronto.edu/~kriz/cifar-10-sample/airplane2.p..."}, {"uri": "https://www.cs.toronto.edu/~kriz/cifar-10-sample/airplane3.p..."}, {"uri": "https://www.cs.toronto.edu/~kriz/cifar-10-sample/airplane4.p..."}, {"uri": "https://www.cs.toronto.edu/~kriz/cifar-10-sample/airplane5.p..."}], "parameters": {"budget": 3}, "execEndpoint":"/query"}'