Protocols and Structures for Inference: A RESTful API for Machine Learning

Journal of Machine Learning Research
Publication Type:
Conference Proceeding
Proceedings of The 2nd International Conference on Predictive APIs and Apps, 2016, pp. 29 - 42
Issue Date:
Full metadata record
Diversity in machine learning APIs (in both software toolkits and web services), works against realising machine learning’s full potential, making it difficult to draw on individual algorithms from different products or to compose multiple algorithms to solve complex tasks. This paper introduces the Protocols and Structures for Inference (PSI) service architecture and specification, which presents inferential entities—relations, attributes, learners and predictors—as RESTful web resources that are accessible via a common but flexible and extensible interface. Resources describe the data they ingest or emit using a variant of the JSON schema language, and the API has mechanisms to support non-JSON data and future extension of service features.
Please use this identifier to cite or link to this item: