WP-SGD: Weighted parallel SGD for distributed unbalanced-workload training system.
Daning ChengShigang LiYunquan ZhangPublished in: J. Parallel Distributed Comput. (2020)
Keyphrases
- stochastic gradient descent
- loss function
- least squares
- training speed
- step size
- random forests
- matrix factorization
- distributed systems
- load balance
- distributed processing
- parallel database systems
- support vector machine
- regularization parameter
- parallel processing
- distributed environment
- parallel execution
- parallel computing
- master slave
- training set
- online algorithms
- importance sampling
- multiple kernel learning
- convergence speed
- fault tolerant
- computer networks
- mobile agents
- response time
- multi agent
- training process
- decision trees
- weight vector
- computing environments
- parallel computation
- filter bank
- supervised learning
- low overhead
- pairwise