feature learning
Feature learning refers to the process through which machine learning models automatically discover the essential characteristics (or features) from raw input data. This is often accomplished through methods such as deep learning, where models learn hierarchical feature representations from complex data.
- Abstain Mask Retain Core: Time Series Prediction by Adaptive Masking Loss with Representation Consistency
- AlphaDecay: Module-wise Weight Decay for Heavy-Tailed Balancing in LLMs
- AltLoRA: Towards Better Gradient Approximation in Low-Rank Adaptation with Alternating Projections
- Alternating Gradient Flows: A Theory of Feature Learning in Two-layer Neural Networks
- ChA-MAEViT: Unifying Channel-Aware Masked Autoencoders and Multi-Channel Vision Transformers for Improved Cross-Channel Learning
- Delving into Cascaded Instability: A Lipschitz Continuity View on Image Restoration and Object Detection Synergy
- Demystifying Spectral Feature Learning for Instrumental Variable Regression
- Doodle to Detect: A Goofy but Powerful Approach to Skeleton-based Hand Gesture Recognition
- Dynamical Decoupling of Generalization and Overfitting in Large Two-Layer Networks
- Dynamical Decoupling of Generalization and Overfitting in Large Two-Layer Networks
- How Data Mixing Shapes In-Context Learning: Asymptotic Equivalence for Transformers with MLPs
- Measuring and Controlling Solution Degeneracy across Task-Trained Recurrent Neural Networks
- Provably Efficient Multi-Task Meta Bandit Learning via Shared Representations
- Rethinking Nighttime Image Deraining via Learnable Color Space Transformation
- Revisiting Residual Connections: Orthogonal Updates for Stable and Efficient Deep Networks
- SegGraph: Leveraging Graphs of SAM Segments for Few-Shot 3D Part Segmentation
- T-REGS: Minimum Spanning Tree Regularization for Self-Supervised Learning
- TS-MOF: Two-Stage Multi-Objective Fine-tuning for Long-Tailed Recognition
- The Computational Advantage of Depth in Learning High-Dimensional Hierarchical Targets
- Towards Multiscale Graph-based Protein Learning with Geometric Secondary Structural Motifs
- Understanding the Evolution of the Neural Tangent Kernel at the Edge of Stability