A Novel Biologically-inspired Visual Cognition Model - Automatic Extraction of Semantics, Formation of Integrated Concepts and Re-selection Features for Ambiguity

CASIA OpenIR > 多模态人工智能系统全国重点实验室 > 机器人理论与应用

	A Novel Biologically-inspired Visual Cognition Model - Automatic Extraction of Semantics, Formation of Integrated Concepts and Re-selection Features for Ambiguity
	P. Yin; H. Qiao; W. Wu; L. Qi; Y. L. Li; S. Zhong; B. Zhang
发表期刊	IEEE Transactions on Cognitive and Developmental Systems
	2017
卷号	PP 期号:99 页码:1-1
摘要	Techniques that integrate neuroscience and information science benefit both fields. Many related models have been proposed in computer vision; however, in general, the robustness and recognition precision are still key problems in object recognition models. In this paper, inspired by the process by which humans recognize objects and its biological mechanisms, a new integrated and dynamic framework is proposed that mimics the semantic extraction, concept formation and feature re-selection found in human visual processing. The main contributions of the proposed model are as follows: (1) Semantic feature extraction: Local semantic features are learned from episodic features extracted from raw images using a deep neural network; (2) Integrated concept formation: Concepts are formed using the local semantic information and structural information is learned through a network; (3) Feature re-selection: When ambiguity is detected during the recognition process, distinctive features based on the differences between the ambiguous candidates are re-selected for recognition. Experimental results on four datasets show that—compared with other methods—the new proposed model is more robust and achieves higher precision for visual recognition, especially when the input samples are semantically ambiguous. Meanwhile, the introduced biological mechanisms further strengthen the interaction between neuroscience and information science.
关键词	Biologically Inspired Model Object Recognition Semantic Learning Structural Learning
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/19891
专题	多模态人工智能系统全国重点实验室_机器人理论与应用
推荐引用方式 GB/T 7714	P. Yin,H. Qiao,W. Wu,et al. A Novel Biologically-inspired Visual Cognition Model - Automatic Extraction of Semantics, Formation of Integrated Concepts and Re-selection Features for Ambiguity[J]. IEEE Transactions on Cognitive and Developmental Systems,2017,PP(99):1-1.
APA	P. Yin.,H. Qiao.,W. Wu.,L. Qi.,Y. L. Li.,...&B. Zhang.(2017).A Novel Biologically-inspired Visual Cognition Model - Automatic Extraction of Semantics, Formation of Integrated Concepts and Re-selection Features for Ambiguity.IEEE Transactions on Cognitive and Developmental Systems,PP(99),1-1.
MLA	P. Yin,et al."A Novel Biologically-inspired Visual Cognition Model - Automatic Extraction of Semantics, Formation of Integrated Concepts and Re-selection Features for Ambiguity".IEEE Transactions on Cognitive and Developmental Systems PP.99(2017):1-1.