基于自适应模糊递归随机配置网络的有限样本食品质量预测研究

《Processes》：An Adaptive Fuzzy Recurrent Stochastic Configuration Network for Food Quality Prediction with Limited Samples

　　

。

摘要

当标记样本有限，且可测量特征与目标质量属性之间的关系是非线性和异质的时，准确预测食品质量仍然具有挑战性。为了解决这个问题，本研究提出了一种自适应模糊递归随机配置网络，该网络整合了自适应模糊规则数量选择、规则耦合的递归子 reservoir、全局残差引导的随机配置以及规则化的分析输出权重学习。该框架在 Mackey–Glass 基准测试和两个实际食品质量预测任务上进行了评估，这两个任务分别涉及荔枝的可溶性糖含量和葡萄酒的质量。在 Mackey–Glass 基准测试中，该网络的均方根误差为 4 period 47 times 10 to the negative 4 power' data-speech-attached="true" data-semantic-braille="??????????????" data-braille-attached="true" has-speech="true"> 4 period 47 times 10 to the negative 4 power' data-semantic-summary ' period separated list of length 2' data-semantic-braille="??????????????"> 4' data-semantic-summary ' number' data-semantic-braille="??">4 period' data-semantic-summary ' period' data-semantic-braille="??">. times times 10 to the negative 4 power' data-semantic-summary ' product with 2 factors' data-semantic-braille="??????????"> 47' data-semantic-summary ' number' data-semantic-braille="???">47 times' data-semantic-summary' times' data-semantic-braille="??"> 10 to the negative 4 power' data-semantic-summary' power' data-semantic-braille="?????"> 10' data-semantic-prefix=' Base' data-semantic-summary '10 negative 4' data-semantic-prefix=' Exponent' data-semantic-summary ' negative number' data-semantic-braille="??"> minus' data-semantic-summary ' minus' data-semantic-braille="?">? 4' data-semantic-summary ' number' data-semantic-braille="?">4, 表示相对于传统模糊递归随机配置网络的 37 period 8 percent' data-speech-attached="true" data-semantic-braille="?????????" data-braille-attached="true" has-speech="true"> 37 period 8 percent' data-semantic-summary ' period separated list of length 2' data-semantic-braille="?????????"> 37' data-semantic-summary ' number' data-semantic-braille="???">37 period' data-semantic-summary ' period' data-semantic-braille="??">. 8 percent' data-semantic-summary ' number percent' data-semantic-braille="????"> 8' data-semantic-summary ' number' data-semantic-braille="??">8