Planning and acting in partially observable stochastic domains
Artificial Intelligence
Scheduling Algorithms for Multiprogramming in a Hard-Real-Time Environment
Journal of the ACM (JACM)
ESP: a system utilization benchmark
Proceedings of the 2000 ACM/IEEE conference on Supercomputing
Reinforcement Learning
Neuro-Dynamic Programming
Scheduling dependent real-time activities
Scheduling dependent real-time activities
HPC Productivity: An Overarching View
International Journal of High Performance Computing Applications
High Performance Computing Productivity Model Synthesis
International Journal of High Performance Computing Applications
Adaptive grid job scheduling with genetic algorithms
Future Generation Computer Systems
Utility accrual real-time scheduling: models and algorithms
Utility accrual real-time scheduling: models and algorithms
Reinforcement learning: a survey
Journal of Artificial Intelligence Research
Brief paper: Average cost temporal-difference learning
Automatica (Journal of IFAC)
Dynamic adaptation of user migration policies in distributed virtual environments
Dynamic adaptation of user migration policies in distributed virtual environments
Adaptive data-aware utility-based scheduling in resource-constrained systems
Journal of Parallel and Distributed Computing
Reinforcement learning ramp metering without complete information
Journal of Control Science and Engineering
| Hi-index | 0.00 |
This paper presents a general methodology for online scheduling of parallel jobs onto multi-processor servers in a soft real-time environment, where the final utility of each job decreases with the job completion time. A solution approach is presented where each server uses Reinforcement Learning for tuning its own value function, which predicts the average future utility per time step obtained from completed jobs based on the dynamically observed state information. The server then selects jobs from its job queue, possibly preempting some currently running jobs and ''squeezing'' some jobs into fewer CPUs than they ideally require to maximize the value of the resulting server state. The experimental results demonstrate the feasibility and benefits of the proposed approach.