Skip to main content

Graphical Inference with Convolutional Neural Networks

Presented by:
Elliot Pickens (Carleton College)
Abstract:

Understanding and recognizing trends in scatter plots is a keep step in many statistical analyses, but these trends are not always obviously apparent. Unclear trends can be particularly problematic during exploratory data analysis. In this paper I present a way to use convolutional neural networks to detect trends in scatter plots, taking some inspiration from previous work done in quantifying scatter plots using scatter plot diagnostics or scagnostics.