Mixed-precision Supernet Training from Vision Foundation Models using Low Rank Adapter.
Yuiko SakumaMasakazu YoshimuraJunji OtsukaAtsushi IrieTakeshi OhashiPublished in: CoRR (2024)
Keyphrases
- low rank
- matrix factorization
- linear combination
- missing data
- neural network
- convex optimization
- image processing
- computer vision
- decision trees
- low rank matrices
- factorization methods
- high dimensional data
- low rank matrix recovery
- robust principal component analysis
- trace norm
- matrix completion
- singular values
- structured prediction
- high order
- probabilistic model
- optical flow
- training set
- feature extraction