convex combination
A mixture of two or more elements (such as model outputs) where the coefficients sum to one, often used in model blending or ensemble methods to combine predictions.
- Enforcing Hard Linear Constraints in Deep Learning Models with Decision Rules
- Explainable Reinforcement Learning from Human Feedback to Improve Alignment
- Principled Model Routing for Unknown Mixtures of Source Domains
- Train with Perturbation, Infer after Merging: A Two-Stage Framework for Continual Learning