Sorting Through ML Algorithms: A Call for Community Contributions

Recent algorithmic advances and the increasing availability of domain data have led to the proliferation of ML (Machine Learning) solutions in a variety of application domains. Domain experts from all fields are called upon, alone or working with a data scientist, to explore using ML techniques to solve their problems. Starting from a domain problem/question, ML-based problem-solving typically involves: (1) formulating the business problem (question) as a data analysis problem (question), (2) sketching the general “workflow” of a ML-based solution, given the needs of the domain and the properties of the available data, and (3) designing and refining the different components of the “workflow” based on the previous two steps, and experimentation. There has to be a consensual body of knowledge that ML researchers agree on, and that ML practitioners can apply to solve the most common problems, regarding the last two steps: (1) the general outline of a ML-based solution, given a data analysis problem, domain requirements, and data characteristics and (2) the selection or tuning of the various components of the solution. Our work deals with capturing this consensual knowledge, and embodying it in a ML problem solving workbench that helps domain specialists who are not ML experts to explore the ML solution space. This paper is concerned with the consensual algorithmic and mathematical properties of data analysis and ML algorithms that can be plugged in ML solution workflows.