针对深度事件检测模型对复杂时序事件检测准确性不足和忽略了不同事件间相关性的问题,提出一种基于信号时态逻辑的深度时序事件检测算法DSTL(Deep Signal Temporal Logic)。该算法一方面引入信号时态逻辑框架,并用信号时态逻辑(STL)公...针对深度事件检测模型对复杂时序事件检测准确性不足和忽略了不同事件间相关性的问题,提出一种基于信号时态逻辑的深度时序事件检测算法DSTL(Deep Signal Temporal Logic)。该算法一方面引入信号时态逻辑框架,并用信号时态逻辑(STL)公式建模时间序列中的事件来综合考虑时间序列上事件的逻辑性和时态性;另一方面采用基于神经网络的基础分类器来检测原子事件的发生情况,并通过STL公式结构和语义来辅助检测复杂事件。另外,使用神经网络模块替代相应的逻辑连接词和时态逻辑算子,从而提供可GPU加速和梯度下降的神经网络模块。通过对6个时间序列数据集的实验,验证了该算法在时序事件检测方面的有效性,并把使用DSTL算法的模型与不使用该算法而使用多层感知机(MLP)、长短期记忆(LSTM)网络和Transformer的深度时间序列分类模型进行比较。实验结果表明,使用DSTL算法的模型在5种事件上的平均F1分数提升了约12%,其中3种跨时间点事件上的平均F1分数提升了约14%,且具备更好的可解释性。展开更多
A method to model and analyze the hybrid systems is presented. The time to be considered in the plant is taken as an explicit parameter through the constrained predicated net (CPN). The CPN's basic structure is a ...A method to model and analyze the hybrid systems is presented. The time to be considered in the plant is taken as an explicit parameter through the constrained predicated net (CPN). The CPN's basic structure is a Petri net with predicated transition. All components of the net are expressed by annotation which is defined on rational set Q. The analysis method for the plant is interval temporal logic represented by Petri nets. This paper combines the above two methods to synthesize the hybrid system, gives a simple and clear expression of the expected action of the studied plant.展开更多
文摘针对深度事件检测模型对复杂时序事件检测准确性不足和忽略了不同事件间相关性的问题,提出一种基于信号时态逻辑的深度时序事件检测算法DSTL(Deep Signal Temporal Logic)。该算法一方面引入信号时态逻辑框架,并用信号时态逻辑(STL)公式建模时间序列中的事件来综合考虑时间序列上事件的逻辑性和时态性;另一方面采用基于神经网络的基础分类器来检测原子事件的发生情况,并通过STL公式结构和语义来辅助检测复杂事件。另外,使用神经网络模块替代相应的逻辑连接词和时态逻辑算子,从而提供可GPU加速和梯度下降的神经网络模块。通过对6个时间序列数据集的实验,验证了该算法在时序事件检测方面的有效性,并把使用DSTL算法的模型与不使用该算法而使用多层感知机(MLP)、长短期记忆(LSTM)网络和Transformer的深度时间序列分类模型进行比较。实验结果表明,使用DSTL算法的模型在5种事件上的平均F1分数提升了约12%,其中3种跨时间点事件上的平均F1分数提升了约14%,且具备更好的可解释性。
文摘A method to model and analyze the hybrid systems is presented. The time to be considered in the plant is taken as an explicit parameter through the constrained predicated net (CPN). The CPN's basic structure is a Petri net with predicated transition. All components of the net are expressed by annotation which is defined on rational set Q. The analysis method for the plant is interval temporal logic represented by Petri nets. This paper combines the above two methods to synthesize the hybrid system, gives a simple and clear expression of the expected action of the studied plant.
