Fully asynchronous stochastic coordinate descent: a tight lower bound on the parallelism achieving linear speedup.
Yun Kuen CheungRichard ColeYixin TaoPublished in: Math. Program. (2021)
Keyphrases
- lower bound
- upper bound
- regret bounds
- lower and upper bounds
- worst case
- branch and bound
- objective function
- branch and bound algorithm
- np hard
- optimal solution
- linear programming relaxation
- parallel processing
- orders of magnitude
- shift register
- parallel execution
- logistic regression
- parallel computing
- sample complexity
- data structure
- stochastic programming
- bayesian networks
- feature selection