ATM Scheduling with Queuing Dely Predictions
Abstract
High speed packet switched networks provide numerous challenges for machine learning based control. In this work we address the problem of scheduling the transmission of packets in a system supporting multiple traffic types with different delay and loss requirements. We develop an adaptive transmission scheduling algorithm, urgency scheduling, that combines elements of memory based function approximation, reinforcement learning and dynamic programming. The resulting learning algorithm should be of more general interest, especially for problems with continuous state spaces that are sensitive to rare events.
Cite
Text
Schwartz. "ATM Scheduling with Queuing Dely Predictions." International Conference on Machine Learning, 1993. doi:10.1016/B978-1-55860-307-3.50046-0Markdown
[Schwartz. "ATM Scheduling with Queuing Dely Predictions." International Conference on Machine Learning, 1993.](https://mlanthology.org/icml/1993/schwartz1993icml-atm/) doi:10.1016/B978-1-55860-307-3.50046-0BibTeX
@inproceedings{schwartz1993icml-atm,
title = {{ATM Scheduling with Queuing Dely Predictions}},
author = {Schwartz, Daniel B.},
booktitle = {International Conference on Machine Learning},
year = {1993},
pages = {306-313},
doi = {10.1016/B978-1-55860-307-3.50046-0},
url = {https://mlanthology.org/icml/1993/schwartz1993icml-atm/}
}