CCF-Based Awareness in Agent Model ABGP

• Published in: Artificial Intelligence in Theory and Practice IV pp. 98 - 107
• Main Authors: Shi, Zhongzhi, Yue, Jinpeng, Ma, Gang, Yang, Xi
• Format: Book Chapter
• Language: English
• Published: Cham Springer International Publishing 2015
• Series: IFIP Advances in Information and Communication Technology
• Subjects: Agent model ABGP; Awareness; Compact coding
• ISBN: 9783319252605; 3319252607
• ISSN: 1868-4238; 1868-422X; 1868-422X
• DOI: 10.1007/978-3-319-25261-2_9

ABGP is an agent cognitive model, which combines external perception and internal mental state of agents together. The model consists of 4-tuple <Awareness, Belief, Goal, Plan>. For environment awareness a framework named Coding and Combining Features (CCF) is proposed in this paper. The CCF framework combines feature descriptors together based on the compact coding. In the CCF framework firstly compact codes are learned to be robust to complex image content variations. Secondly, the compact codes from multiple information sources are combined by multi-kernel learning to generate the final feature. The experiment results show us that CCF method achieves a better performance in the identification task.