ensemble methods
Ensemble methods are techniques that combine multiple models to enhance predictive performance and robustness. By aggregating the outputs of diverse models, such as through voting or averaging, ensemble methods can help mitigate individual model biases and reduce error rates.
- Anomaly Detection by an Ensemble of Random Pairs of Hyperspheres
- Asymmetric Duos: Sidekicks Improve Uncertainty
- Conformal Prediction for Ensembles: Improving Efficiency via Score-Based Aggregation
- FreeInv: Free Lunch for Improving DDIM Inversion
- LASeR: Learning to Adaptively Select Reward Models with Multi-Arm Bandits
- MOOSE-Chem2: Exploring LLM Limits in Fine-Grained Scientific Hypothesis Discovery via Hierarchical Search
- Noise Matters: Optimizing Matching Noise for Diffusion Classifiers
- SpecEM: Training-Free LLM Ensembling via Iterative Drafting, Verification, and Online Feedback
- Uncertainty-Based Smooth Policy Regularisation for Reinforcement Learning with Few Demonstrations