Pretrained neural policies replace repeated GP inference and constrained optimization in safe active learning with a single forward pass, yielding large speedups while preserving query quality.
The orthogonal matrix Q is obtained by performing a QR- decomposition of a sampled A ∈ RD×D (each entity from Uniform[−1, 1])
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Amortized Safe Active Learning for Real-Time Data Acquisition: Pretrained Neural Policies From Simulated Nonparametric Functions
Pretrained neural policies replace repeated GP inference and constrained optimization in safe active learning with a single forward pass, yielding large speedups while preserving query quality.