Flow-Based Programming for Machine Learning

Machine Learning (ML) has gained prominence and tremendous applications in fields like medicine, biology, geography astrophysics, to name a few. Arguably, such areas, it is used by domain experts, who are not necessarily skilled-programmers. Thus, presents steep learning curve for experts programming ML applications. To overcome this foster widespread adoption of techniques, we propose equip them with domain-specific graphical tools. Such tools, based on the principles flow-based paradigm, would support composition at higher level abstraction auto-generation target code. Accordingly, (i) have modelled algorithms as composable components; (ii) described an approach parse flow created connecting several components use API-based code generation technique generate application. demonstrate feasibility our conceptual approach, APIs Apache Spark validated three use-cases. The use-cases designed capture ease program specification level, easy parametrisation APIs, application auto-validation generated model better prediction accuracy.