Summary of Skippredict: When to Invest in Predictions For Scheduling, by Rana Shahout et al.
SkipPredict: When to Invest in Predictions for Scheduling
by Rana Shahout, Michael Mitzenmacher
First submitted to arxiv on: 5 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Distributed, Parallel, and Cluster Computing (cs.DC)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper addresses scheduling in queueing systems where predictions of job sizes are used to optimize resource allocation. However, existing work assumes that these predictions are cost-free, which is unrealistic. To account for the cost of predictions, the authors introduce SkipPredict, a novel approach that categorizes jobs based on their prediction requirements. This approach employs cheap one-bit predictions to classify jobs as short or long and then applies more detailed predictions for longer jobs. The analysis considers two models: external cost model, where prediction generation incurs a cost but does not affect job service times; and server time cost model, where predictions require server processing time and are scheduled alongside the jobs. This paper aims to provide a more realistic representation of the scheduling process by accounting for the cost of predictions. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper talks about how to make decisions in a system that deals with jobs and resources. Right now, people assume that knowing the size of each job is free, but it’s not. The authors are trying to figure out how to balance this cost against getting more accurate information. They came up with a new way called SkipPredict that looks at each job and says whether it needs simple or detailed predictions. This helps prioritize jobs in the most efficient way possible. There are two ways they’re thinking about this: one where making predictions takes some time, but doesn’t affect how long each job takes; and another where making predictions also uses up some of the same resources as the jobs themselves. |
