2 papers
cs.LG2026
Neighborhood Smoothing for Calibration
Idan Horowitz, Avigdor Gal
Modern neural networks are often miscalibrated, with a tendency to overconfidence. Existing train-time calibration methods largely modify task losses or calibration penalties, leav…
cs.LG2026
Train Smarter, Not Harder: Switching Signal-Guided Training in Active Learning
Nagham Omar, Maya Rozenshtein, Evgeny Mishlyakov +1
Training strategy, namely whether to retrain from scratch or fine-tune from the previous checkpoint, is an overlooked decision variable in active learning. We show that this choice…