Floating Point Architecture Extensions for Optimized Matrix Factorization.
Ardavan PedramAndreas GerstlauerRobert A. van de GeijnPublished in: IEEE Symposium on Computer Arithmetic (2013)
Keyphrases
- matrix factorization
- floating point
- floating point arithmetic
- instruction set
- collaborative filtering
- recommender systems
- low rank
- missing data
- negative matrix factorization
- fixed point
- factor analysis
- data sparsity
- nonnegative matrix factorization
- floating point unit
- sparse matrices
- data matrix
- item recommendation
- probabilistic matrix factorization
- factorization methods
- face recognition
- personalized ranking
- markov random field
- stochastic gradient descent