Towards Human-In-The-Loop Data Exploration

Machine learning (ML) and advanced statistics are important tools for drawing insights from large datasets. However, these techniques often require human intervention to steer computation towards meaningful results. In this talk, I will present early results on Vizdom, a new human-in-the-loop (HIL) system for interactive analytics. Vizdom's has a novel touch and pen front-end that allows users to visually compose complex workflows of ML and statistics operators on an interactive whiteboard, whereas the backend leverages recent advances in workflow compilation techniques developed at Brown University to run these computations at interactive speeds.