Ali Fakhri Mahdi Al-Jumaily, Abdulmajeed Al-Jumaily, Saba Jasim Al-Jumaili
الملخص الانجليزي
Most of the existing deep learning-based business process remaining time prediction methods use traditional long-short-term memory recurrent neural networks to build prediction models. Due to the limited modeling ability of traditional long-short-term memory recurrent neural networks for sequence data, and existing methods there is still much room for improvement in the prediction effect. Aiming at the shortcomings of existing methods, this paper proposes a business process remaining time prediction method based on attention bidirectional recurrent neural network. The method uses a bidirectional recurrent neural network to model the process instance data and introduces an attention mechanism to automatically learn the weights of different events in the process instance. In addition, in order to further improve the learning effect, an iterative learning strategy is designed based on the idea of transfer learning, which builds remaining time prediction models for process instances of different lengths, which improves the pertinence of the model. The experimental results show that the proposed method has obvious advantages compared with traditional methods.
علي فخري مهدي الجميلي | ALI FAKHRI MAHDI | 1600