out-of-distribution generalization
The ability of a machine learning model to perform well on data that differs from the training distribution. This is crucial for real-world application where models encounter unseen or novel inputs.
- Aggregation Hides Out-of-Distribution Generalization Failures from Spurious Correlations
- Among Us: A Sandbox for Measuring and Detecting Agentic Deception
- Brain-like Variational Inference
- Causally Reliable Concept Bottleneck Models
- Effective Neural Approximations for Geometric Optimization Problems
- FAPEX: Fractional Amplitude-Phase Expressor for Robust Cross-Subject Seizure Prediction
- From Pretraining to Pathology: How Noise Leads to Catastrophic Inheritance in Medical Models
- Generative RLHF-V: Learning Principles from Multi-modal Human Preference
- LORE: Lagrangian-Optimized Robust Embeddings for Visual Encoders
- Learning a Cross-Modal Schrödinger Bridge for Visual Domain Generalization
- MEMOIR: Lifelong Model Editing with Minimal Overwrite and Informed Retention for LLMs
- ML4CFD Competition: Results and Retrospective Analysis
- Measure-Theoretic Anti-Causal Representation Learning
- OMEGA: Can LLMs Reason Outside the Box in Math? Evaluating Exploratory, Compositional, and Transformative Generalization
- Pruning Spurious Subgraphs for Graph Out-of-Distribution Generalization
- Quantifying Distributional Invariance in Causal Subgraph for IRM-Free Graph Generalization
- SpiderSolver: A Geometry-Aware Transformer for Solving PDEs on Complex Geometries
- Tropical Attention: Neural Algorithmic Reasoning for Combinatorial Algorithms
- Vicinal Label Supervision for Reliable Aleatoric and Epistemic Uncertainty Estimation