On the parallel I/O optimality of linear algebra kernels: near-optimal matrix factorizations.
Grzegorz KwasniewskiMarko KabicTal Ben-NunAlexandros Nikolaos ZiogasJens Eirik SaethreAndré GaillardTimo SchneiderMaciej BestaAnton KozhevnikovJoost VandeVondeleTorsten HoeflerPublished in: SC (2021)
Keyphrases
- linear algebra
- matrix factorization
- collaborative filtering
- tensor factorization
- low rank
- recommender systems
- negative matrix factorization
- missing data
- singular value decomposition
- kernel function
- nonnegative matrix factorization
- computer architecture
- image processing
- feature space
- parallel computers
- matrix multiplication
- binary matrix
- sparse matrices
- multiple kernel learning
- natural images
- data processing
- data points
- data analysis
- face recognition
- image segmentation