• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Conference on Computer and Information Sciences >Ridge Polynomial Neural Network with Error Feedback for Recursive Multi-step Forecast Strategy: A Case Study of Carbon Dioxide Emissions Forecasting
【24h】

Ridge Polynomial Neural Network with Error Feedback for Recursive Multi-step Forecast Strategy: A Case Study of Carbon Dioxide Emissions Forecasting

机译：具有误差反馈的Ridge多项式神经网络的递归多步预测策略：以二氧化碳排放量预测为例

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Neural networks (NNs) have been used extensively for forecasting problems. NN with error feedbacks is a type of NNs that showed more accurate forecasts compared to feedforward NNs and NNs with output feedbacks with some forecasting problems. The main issue with NN s with error feedbacks appears when there is a need for recursive multi-step forecast strategy because the observed values must be known in order to calculate network errors. This paper proposes to use the last calculated error after finishing training NNs with error feedbacks because the observed values are unknown. This last calculated error is used as a fixed value when producing forecasts using recursive multi-step forecast strategy. For that, this paper investigated this simple solution with a NN with error feedback called the ridge polynomial neural network with error feedback (RPNN-EF). Carbon dioxide emissions for three countries in the organization of the petroleum exporting countries (OPEC) were used in this investigation. The forecasting accuracy of RPNN-EF was compared with seven forecasting methods. According to the obtained results, on average, the proposed solution produces reasonable forecasts compared to the seven forecasting methods. Therefore, this solution can be suggested for NNs with error feedbacks for recursive multi-step forecast strategy.
机译：神经网络（NNs）已广泛用于预测问题。带有错误反馈的NN是一种神经网络，与前馈NN和具有输出反馈且有一些预测问题的NN相比，具有更准确的预测。当需要递归多步预测策略时，具有误差反馈的NN的主要问题就会出现，因为必须知道观测值才能计算网络误差。本文建议在完成带有误差反馈的训练神经网络后使用最后计算出的误差，因为观测值未知。使用递归多步预测策略生成预测时，将最后计算出的误差用作固定值。为此，本文研究了带有错误反馈的神经网络的简单解决方案，称为带有错误反馈的岭多项式神经网络（RPNN-EF）。本次调查使用了石油输出国组织（OPEC）中三个国家的二氧化碳排放量。将RPNN-EF的预测精度与7种预测方法进行了比较。根据获得的结果，与七种预测方法相比，所提出的解决方案平均可以产生合理的预测。因此，可以为具有错误反馈的神经网络建议该解决方案，以进行递归多步预测策略。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号