MECHVAR: Variance-Guided Mechanism Discrimination for Autonomous Machine Learning Experiment Selection

Excerpt

arXiv:2610.01819v1 Announce Type: new Abstract: Benchmark gains are often mechanism-ambiguous: reproducing an improvement does not by itself identify why it occurs. We study finite-library mechanism discrimination, where posterior-weighted candidate mechanisms, executable probes, and a limited experimental budget define a sequential experiment-selection problem. MECHVAR selects the next probe by maximizing the posterior-weighted variance of its predicted responses. Under a shared-Gaussian predic