mutual information
A measure of the amount of information one random variable contains about another. In AI, it is often used to quantify the relationship between features or distributions, aiding in feature selection and understanding model behavior.
- Activation-Guided Consensus Merging for Large Language Models
- Adaptive Gradient Masking for Balancing ID and MLLM-based Representations in Recommendation
- Aligning Text to Image in Diffusion Models is Easier Than You Think
- Bits Leaked per Query: Information-Theoretic Bounds for Adversarial Attacks on LLMs
- Breaking AR’s Sampling Bottleneck: Provable Acceleration via Diffusion Language Models
- Composite Flow Matching for Reinforcement Learning with Shifted-Dynamics Data
- Conformal Information Pursuit for Interactively Guiding Large Language Models
- Connecting Jensen–Shannon and Kullback–Leibler Divergences: A New Bound for Representation Learning
- Decomposing stimulus-specific sensory neural information via diffusion models
- Demystifying Reasoning Dynamics with Mutual Information: Thinking Tokens are Information Peaks in LLM Reasoning
- Each Complexity Deserves a Pruning Policy
- Enhancing Privacy in Multimodal Federated Learning with Information Theory
- Evolution of Information in Interactive Decision Making: A Case Study for Multi-Armed Bandits
- How Patterns Dictate Learnability in Sequential Data
- Incomplete Multi-view Clustering via Hierarchical Semantic Alignment and Cooperative Completion
- Information-Driven Design of Imaging Systems
- Information-Theoretic Discrete Diffusion
- Learning Task-Agnostic Representations through Multi-Teacher Distillation
- MI-TRQR: Mutual Information-Based Temporal Redundancy Quantification and Reduction for Energy-Efficient Spiking Neural Networks
- Missing Data Imputation by Reducing Mutual Information with Rectified Flows
- Mixture-of-Experts Meets In-Context Reinforcement Learning
- Multi-modal contrastive learning adapts to intrinsic dimensions of shared latent variables
- Neural Mutual Information Estimation with Vector Copulas
- TRiCo: Triadic Game-Theoretic Co-Training for Robust Semi-Supervised Learning
- Towards Building Model/Prompt-Transferable Attackers against Large Vision-Language Models