Reproduce “STUDD: A Student-Teacher Method for Unsupervised Concept Drift Detection” 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
Vitor Cerqueira, Heitor Murilo Gomes, Albert Bifet, Luis Torgo
STUDD monitors concept drift in changing data streams without requiring true labels after initial training. It trains a student to imitate the deployed teacher, then sends prediction disagreement to a drift detector. Across 19 benchmark streams, it had an average rank of 3.50 for label cost but 6.97 for predictive performance, indicating a label-cost trade-off rather than uniformly better detection.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.