Multi-Step-Ahead Web Service QoS Time Series Forecasting: A Multi-Predictor-Based Genetic Programming Approach

Created by W.Langdon from gp-bibliography.bib Revision:1.7644

@InProceedings{syu:2023:GECCOcomp,

• author = "Yang Syu and Yong-Yi Fanjiang",
• title = "Multi-Step-Ahead Web Service {QoS} Time Series Forecasting: A Multi-Predictor-Based Genetic Programming Approach",
• booktitle = "Proceedings of the 2023 Genetic and Evolutionary Computation Conference",
• year = "2023",
• editor = "Alberto Moraglio",
• pages = "43--44",
• address = "Lisbon, Portugal",
• series = "GECCO '23",
• month = "15-19 " # jul,
• organisation = "SIGEVO",
• publisher = "Association for Computing Machinery",
• publisher_address = "New York, NY, USA",
• keywords = "genetic algorithms, genetic programming, web services, machine learning, time series forecasting, service-oriented software engineering",
• isbn13 = "9798400701191",
• DOI = "doi:10.1145/3583133.3595841",
• size = "2 pages",
• abstract = "Previously, in a GECCO 2022 Hot-off-the-Press paper [1], we presented a comprehensive survey of the modeling and prediction of Web service (WS) quality of service (QoS) time series [2]. Based on the exhaustive investigation in [2], this research subject has already been deeply and widely studied for over a decade; for the one-step-ahead version of this problem, which can be considered its most primitive problem form, overall, our proposed and developed genetic programming (GP)-based solution outperforms competitors in terms of both modeling and forecasting accuracy, according to our ongoing study, which has been reported in [3] [4] [5]. Nevertheless, as argued in [6], for the long-term use and rental of cloud-based WSs, multi-step-ahead QoS time series prediction of these services is needed. Thus, the authors employed and revised the two most widely used single-predictor-based time series methods, namely, autoregressive integrated moving average (ARIMA) models and exponential smoothing (ES), to address this latest version of the problem.For this multi-step-ahead variant of the problem, in Y. Y. Fanjiang, Y. Syu and W. L. Huang, {"}Time Series QoS Forecasting for Web Services Using Multi-Predictor-based Genetic Programming{"}, IEEE Transactions on Services Computing (TSC), Vol. 15, P.P. 1423--1435, 2022, we devise and employ a multipredictor-based approach to genetic programming. First, due to its superiority in our past work for the basic (i.e., one-step-ahead) version of the problem, we investigate the performance of GP on this newly emerged (multi-step-ahead) form of the problem. Second, instead of using a single model for predictions regarding multiple future time points, which is the method commonly adopted in prior research [2], we evolve and apply a dedicated GP-generated predictor for each targeted future time point and its projection. Furthermore, two different strategies for the consumed predictor inputs are tested to determine their differences and influence on accuracy so that a better strategy can be empirically determined. In addition, we propose in the reported paper [7] two disparate techniques to further enhance the resulting performance of our multi-predictor-based GP method.As in our previous GECCO Hot-off-the-Press paper [1], this abstract paper presents to the GECCO community a verified application of GP on a more difficult and challenging type of WS QoS time series forecasting. Our purpose is to enable the GECCO community to use this application of GP, to try to improve GP to obtain more accurate and better results, and to investigate other potential evolutionary paradigms and techniques for this issue.",
• notes = "GECCO-2023 A Recombination of the 32nd International Conference on Genetic Algorithms (ICGA) and the 28th Annual Genetic Programming Conference (GP)",

}

Genetic Programming entries for Yang Syu Yong-Yi FanJiang

Citations