March 10th

14:30 , R2014 Digiteo Shannon (660) (see location):

Balasz Kegl

Title : Rapid Analytics and Model Prototyping

Abstract :

An important mission of the Center for Data Science is to develop and manage tools that allow data scientist to use their expertise in en efficient way to solve data analytics problems coming from (domain) scientific projects. The subject of this talk is one of those tools that we baptized "Rapid Analytics and Model Prototyping" (RAMP). I will explain the goals, the format, the rationale behind our design choices, and the social and the technical aspects of the tool.

One of the most interesting outputs of the RAMPs is that they will generate ample data and a playing field for our research project on collaborative hyperparameter optimization. The ultimate goal is to design an environment in which human experts can interact with automatic optimization tools, combining human creativity with robust automatic exploration in a rapid feedback loop.

Although we accommodate micro-contributions of novice data scientists, one key aspect of the RAMPs is that we also need data science experts to drive the exploration, so one of the goals of this presentation is to generate interest within the group.