orthogonal transformations
In linear algebra, orthogonal transformations refer to transformations that preserve distances and angles, typically involving rotation or reflection. In the context of AI, such transformations can be useful in methods like PCA and neural network design to maintain the structure of data during processing.
- $\boldsymbol{\lambda}$-Orthogonality Regularization for Compatible Representation Learning
- Bridging Critical Gaps in Convergent Learning: How Representational Alignment Evolves Across Layers, Training, and Distribution Shifts
- DenoiseRotator: Enhance Pruning Robustness for LLMs via Importance Concentration
- Generalized Linear Mode Connectivity for Transformers
- Generalized Linear Mode Connectivity for Transformers