Reproduce “Modeling Art Evaluations from Comparative Judgments: A Deep Learning Approach to Predicting Aesthetic Preferences” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Manoj Reddy Bethi, Sai Rupa Jhade, Pravallika Yaganti, Monoshiz Mahbub Khan, Zhe Yu
The authors test whether painting preferences can be modeled from pairwise choices instead of numerical ratings. They train a ResNet-50 encoder using preference directions derived from existing ratings, and compare it with rating regression. For average Abstract Beauty, deep regression raises R² from 0.090 to 0.385 versus hand-crafted features. A five-person survey found faster pairwise judgments; benchmark pairs were generated from ratings.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.