Facto-CNN: Memory-Efficient CNN Training with Low-rank Tensor Factorization and Lossy Tensor Compression.
Seungtae LeeJonghwan KoSeokin HongPublished in: ACML (2023)
Keyphrases
- tensor factorization
- low rank
- matrix factorization
- memory efficient
- tensor decomposition
- singular value decomposition
- factorization methods
- collaborative filtering
- negative matrix factorization
- missing data
- recommender systems
- convex optimization
- trace norm
- high order
- nonnegative matrix factorization
- training set
- higher order
- auxiliary information
- factorization method
- least squares
- high dimensional data
- nearest neighbor
- data representation
- principal component analysis
- gaussian processes
- kl divergence
- semi supervised
- active learning
- image processing
- feature extraction