Cross-sample prediction churn between bootstrap-trained classifiers reaches 8-22% on chemistry benchmarks; K-bootstrap bagging reduces it 40-54% and twin-bootstrap with sym-KL consistency loss reduces it a further median 45% at matched 2x compute.
Machine Learning , year =
5 Pith papers cite this work, alongside 16,377 external citations. Polarity classification is still indexing.
5
Pith papers citing it
16.4k
external citations · OpenAlex
years
2026 5representative citing papers
LG-CoTrain, an LLM-guided co-training method, outperforms classical semi-supervised baselines for crisis tweet classification in low-resource settings with 5-25 labeled examples per class.
Operator Boosting constructs compact neural-operator PDE surrogates by sequential residual learning with validation-selected shrinkage, yielding 72-95% parameter reduction and accuracy gains on 21 of 30 dataset-architecture pairs.
Path-based adaptive weighting of random forest trees via decision path patterns delivers statistically significant accuracy gains on 36 binary classification benchmarks with minimal class-recall regression.
Diverse ensembles of prompted and fine-tuned GPT-4.1-Mini monitors achieve 2.4x better detection of flawed code solutions than homogeneous ensembles on adversarial inputs.
citing papers explorer
-
Reducing cross-sample prediction churn in scientific machine learning
Cross-sample prediction churn between bootstrap-trained classifiers reaches 8-22% on chemistry benchmarks; K-bootstrap bagging reduces it 40-54% and twin-bootstrap with sym-KL consistency loss reduces it a further median 45% at matched 2x compute.
-
LLM-guided Semi-Supervised Approaches for Social Media Crisis Data Classification
LG-CoTrain, an LLM-guided co-training method, outperforms classical semi-supervised baselines for crisis tweet classification in low-resource settings with 5-25 labeled examples per class.
-
Operator Boosting Produces Pareto-Efficient PDE Surrogates
Operator Boosting constructs compact neural-operator PDE surrogates by sequential residual learning with validation-selected shrinkage, yielding 72-95% parameter reduction and accuracy gains on 21 of 30 dataset-architecture pairs.
-
Decision-Path Patterns as Tree Reliability Signals: Path-based Adaptive Weighting for Random Forest Classification
Path-based adaptive weighting of random forest trees via decision path patterns delivers statistically significant accuracy gains on 36 binary classification benchmarks with minimal class-recall regression.
-
Ensemble Monitoring for AI Control: Diverse Signals Outweigh More Compute
Diverse ensembles of prompted and fine-tuned GPT-4.1-Mini monitors achieve 2.4x better detection of flawed code solutions than homogeneous ensembles on adversarial inputs.