Snowcat and CAVA: Visualization Tools for Interacting with AutoML and Knowledgebases

The goal of our project is to create visualization systems that enable subject matter experts (SMEs) to construct, curate, evaluate, and assess data-centric machine learning models. Our systems allow a SME to explore data visually, construct objective functions via intuitive interfaces, and submit them to an AutoML system to generate machine learning models. By integrating visual exploration of input data, model outputs, and visualization results, our systems supports model development, tuning, formalization, validation in an intuitive manner decoupled from the underlying modeling techniques.