theoretical justification
The formal reasoning or proofs that support the efficacy and appropriateness of algorithms and models, validating their use in practical applications and fostering confidence in their predictions.
- A Beyond-Worst-Case Analysis of Greedy k-means++
- A Principled Path to Fitted Distributional Evaluation
- Accelerating Optimization via Differentiable Stopping Time
- Conditioning Matters: Training Diffusion Policies is Faster Than You Think
- Fine-Tuning Discrete Diffusion Models with Policy Gradient Methods
- Over-squashing in Spatiotemporal Graph Neural Networks
- Simultaneous Statistical Inference for Off-Policy Evaluation in Reinforcement Learning
- Spectral Analysis of Diffusion Models with Application to Schedule Design
- Tail-Optimized Caching for LLM Inference