DoRA: Weight-Decomposed Low-Rank Adaptation.
Shih-Yang LiuChien-Yi WangHongxu YinPavlo MolchanovYu-Chiang Frank WangKwang-Ting ChengMin-Hung ChenPublished in: CoRR (2024)
Keyphrases
- low rank
- linear combination
- convex optimization
- missing data
- matrix factorization
- singular value decomposition
- low rank matrix
- matrix completion
- rank minimization
- high order
- kernel matrix
- high dimensional data
- matrix decomposition
- minimization problems
- semi supervised
- low rank matrices
- singular values
- trace norm
- robust principal component analysis
- weight vector
- data matrix
- neural network
- affinity matrix
- low rank approximation
- knn
- low rank representation
- computer vision
- machine learning