Transfer learning algorithms can transform prior knowledge into linearization knowledge to model nonlinear systems. However, the linearization knowledge-based models tend to diverge in the process of knowledge lineari...
详细信息
Transfer learning algorithms can transform prior knowledge into linearization knowledge to model nonlinear systems. However, the linearization knowledge-based models tend to diverge in the process of knowledge linearization due to the neglected information of higher-order terms. To overcome this problem, a second-order knowledgefilter transfer learning algorithm(SOFTLA) is developed for modeling nonlinear systems. First, a knowledge transformation strategy is introduced to transform the linearization source knowledge into comprehensive knowledge containing first-order and second-order *** with the original knowledge, the transformed source knowledge with second-order term can prevent information loss during the knowledge linearization. Second, a knowledge filter algorithm is proposed to eliminate the useless information in the source knowledge. Subsequently, a suitable filter gain is designed to reduce the cumulative error in knowledge updating process. Third, a model adaptation mechanism is designed to enable effective knowledge transfer by updating the structure and parameters of the target model simultaneously. Subsequently, the adaptability of the source knowledge is enhanced to facilitate learning tasks in the target domain. Finally, a benchmark problem and several practical industrial applications are presented to validate the superiority of SOFTLA. The experimental discussions illustrate that SOFTLA can obtain obvious advantages over contrastive methods.
Transfer learning algorithms are capable to apply previously learned knowledge in source domain, which alleviates much expensive efforts of knowledge recollection in target domain. But the knowledge in source domain i...
详细信息
Transfer learning algorithms are capable to apply previously learned knowledge in source domain, which alleviates much expensive efforts of knowledge recollection in target domain. But the knowledge in source domain is always imperfect due to redundant or contaminated information. To solve this problem, an ensemble filtertransfer learning (EFTL) algorithm based on the source knowledge reconstruction is proposed in this paper. First, a knowledge partition strategy based on model is developed to classify the source knowledge into different types. Then, the positive knowledge can be identified, which contributes to target domain with a rejection of the negative transfer. Second, a knowledge filter algorithm is introduced to filter out the redundant information in non-positive knowledge. Then, the non-positive knowledge can be reconstructed by this algorithm to prevent the loss of available information. Third, an ensemble transfer mechanism is established to realize the synchronous transfer of omnidirectional knowledge for the target domain. Finally, comparative experiments on model prediction in practical applications are provided to illustrate the dependability of EFTL.
暂无评论