theoretical understanding
Theoretical understanding in AI involves developing principles and frameworks that explain and predict the behavior of models, algorithms, and their interactions. This understanding provides insights that help guide the design and improvement of machine learning systems.
- A Reliable Cryptographic Framework for Empirical Machine Unlearning Evaluation
- Does Stochastic Gradient really succeed for bandits?
- Does Stochastic Gradient really succeed for bandits?
- Greedy Sampling Is Provably Efficient For RLHF
- Information Theoretic Learning for Diffusion Models with Warm Start
- Learning to Generalize: An Information Perspective on Neural Processes
- Length Generalization via Auxiliary Tasks
- On the Edge of Memorization in Diffusion Models
- On the Stability of Graph Convolutional Neural Networks: A Probabilistic Perspective
- REINFORCE Converges to Optimal Policies with Any Learning Rate
- Reasoning by Superposition: A Theoretical Perspective on Chain of Continuous Thought
- Second-order Optimization under Heavy-Tailed Noise: Hessian Clipping and Sample Complexity Limits
- Trained Mamba Emulates Online Gradient Descent in In-Context Linear Regression