Papers
12094 papers
ICML2025
DIME: Diffusion-Based Maximum Entropy Reinforcement Learning
Summary pending...
Reinforcement LearningDiffusion ModelsDiffusion Based Reinforcement Learning
ICML2025
Exploring Large Action Sets with Hyperspherical Embeddings using von Mises-Fisher Sampling
Summary pending...
Efficient SamplingProbability TheoryLarge-Scale
ICML2025
Provable Length Generalization in Sequence Prediction via Spectral Filtering
Summary pending...
sequence prediction; length generalization
ICML2025
AutoCATE: End-to-End, Automated Treatment Effect Estimation
Summary pending...
Treatment Effect EstimationCausal InferenceAutoML
ICML2025
Partially Observable Reinforcement Learning with Memory Traces
Summary pending...
Reinforcement learning theoryPartial observabilityMemory
ICML2025
Improved and Oracle-Efficient Online $\ell_1$-Multicalibration
Summary pending...
online multicalibrationonline learningoracle-efficient algorithms
ICML2025
Active Reward Modeling: Adaptive Preference Labeling for Large Language Model Alignment
Summary pending...
Reward modelingactive learningLLM alignment
ICML2025
Temporal Misalignment in ANN-SNN Conversion and its Mitigation via Probabilistic Spiking Neurons
Summary pending...
Spiking Neural NetworksSNNANN-SNN conversion
ICML2025
SCENIR: Visual Semantic Clarity through Unsupervised Scene Graph Retrieval
Summary pending...
Scene Graph RetrievalUnsupervised Graph AutoencodersVisual Semantic Similarity
ICML2025
How Do Large Language Monkeys Get Their Power (Laws)?
Summary pending...
scaling lawsinference computescaling inference compute
ICML2025
A Geometric Approach to Personalized Recommendation with Set-Theoretic Constraints Using Box Embeddings
Summary pending...
Box EmbeddingsPersonalized QuerySet-based embeddings
ICML2025
Understanding the Logic of Direct Preference Alignment through Logic
Summary pending...
preference learningneuro-symboliclogic
ICML2025
Hierarchical Reinforcement Learning with Uncertainty-Guided Diffusional Subgoals
Summary pending...
Hierarchical Reinforcement Learning
ICML2025
On the Power of Context-Enhanced Learning in LLMs
Summary pending...
In-Context LearningSample EfficiencyKnowledge Internalization
ICML2025
Catoni Contextual Bandits are Robust to Heavy-tailed Rewards
Summary pending...
Heavy-tailed RewardsContextual banditsGenenral function approximation
ICML2025
SWAN: SGD with Normalization and Whitening Enables Stateless LLM Training
Summary pending...
Efficient LLM optimizationStateless Optimizers
ICML2025
Scaling Laws for Forgetting during Finetuning with Pretraining Data Injection
Summary pending...
LLMfine-tuningscaling laws
ICML2025
LLMs on the Line: Data Determines Loss-to-Loss Scaling Laws
Summary pending...
LLMsscaling lawsdata-centric ML
ICML2025
Soup-of-Experts: Pretraining Specialist Models via Parameters Averaging
Summary pending...
Pre-trainingSpecializationDomain Adaptation