An information fusion method based on deep learning and fuzzy discount-weighting for target intention recognition. (March 2022)

• Record Type: Journal Article
• Authors: Zhang, Zhuo; Wang, Hongfei; Geng, Jie; Jiang, Wen; Deng, Xinyang; Miao, Wang
• Is Part Of: Engineering applications of artificial intelligence. Volume 109(2022)
• Publication Date: 2022-03
• Subjects: Information fusion -- Dempster–Shafer Theory -- Fuzzy discount -- Target intention recognition -- Deep learning; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2021.104610
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: In the military confrontation environment, recognizing the target intention is helpful to know the target actions in advance, and the global intention recognition of target formations can provide decision-making reference for the command center. In this work, a new information fusion method for multi-target formation intention recognition is developed, which combines the advantages of deep learning and Dempster–Shafer theory. This method firstly construct the deep learning networks and design the corresponding conversion methods to obtain the uncertain information for target intention recognition. Then, a new fuzzy discount-weighting operation is proposed. This operation defines the new fuzzy discount rule and fuzzy weighting rule, which generates discount evidence and weighting coefficients to improve the reliability of evidence, then obtain a more reasonable fusion result. The simulation results show that the method is effective and feasible for global target intention recognition under uncertain and incomplete information. Highlights: A new information fusion method based on deep learning and fuzzy discount-weighting for target intention recognition is proposed. The proposed method can convert the probability distribution output by deep learning networks into the uncertain information for recognition. The proposed fuzzy discount operation can use the external information to modify the original evidence. The proposed fuzzy weighting operation can combine the external and internal information to improve the reliability of fusion result.