Learning Theory

Theory on Mixture-of-Experts in Continual Learning ICLR 2025 The Perils of Optimizing Learned Reward Functions: Low Training Error Does Not Guarantee Low Regret ICML 2025 The Pitfalls of Imitation Learning when Actions are Continuous COLT 2025 The Reasoning-Memorization Interplay in Language Models Is Mediated by a Single Direction ACL 2025 The Relationship Between No-Regret Learning and Online Conformal Prediction ICML 2025 There’s No Such Thing as Simple Reasoning for LLMs ACL 2025 The Role of Randomness in Stability ICML 2025 The Sample Complexity of Online Strategic Decision Making with Information Asymmetry and Knowledge Transportability ICML 2025 The Sample Complexity of Stackelberg Games AISTATS 2025 The Size of Teachers as a Measure of Data Complexity: PAC-Bayes Excess Risk Bounds and Scaling Laws AISTATS 2025 The Space Complexity of Learning-Unlearning Algorithms (extended abstract) COLT 2025 The Surprising Effectiveness of Infinite-Width NTKs for Characterizing and Improving Model Training AAAI 2025 Three Mechanisms of Feature Learning in a Linear Network ICLR 2025 Tight and Fast Bounds for Multi-Label Learning ICML 2025 TimeInf: Time Series Data Contribution via Influence Functions ICLR 2025 To Code or Not To Code? Exploring Impact of Code in Pre-training ICLR 2025 Tokenization and Representation Biases in Multilingual Models on Dialectal NLP Tasks EMNLP 2025 Topology-Aware Dynamic Reweighting for Distribution Shifts on Graph ICML 2025 Towards Advanced Mathematical Reasoning for LLMs via First-Order Logic Theorem Proving EMNLP 2025 Towards a Holistic and Automated Evaluation Framework for Multi-Level Comprehension of LLMs in Book-Length Contexts EMNLP 2025 Towards a learning theory of representation alignment ICLR 2025 Towards Auto-Regressive Next-Token Prediction: In-context Learning Emerges from Generalization ICLR 2025 Towards Bridging Generalization and Expressivity of Graph Neural Networks ICLR 2025 Towards counterfactual fairness through auxiliary variables ICLR 2025 Towards Data Contamination Detection for Modern Large Language Models: Limitations, Inconsistencies, and Oracle Challenges COLING 2025