The Research Of Topology As A Cognitive Approach

Cognitive science has integrated many sciences.It has been the focus of many researchers to explore the cognitive system of human beings and apply the principle to the artificial system.In recent years,with the development of brain science,human has made great progress in the study of the nervous system.These advances provide important theoretical basis for the realization of simulation in engineering.First,the hierarchical structure of the brain and information processing mechanisms reflect some important topological properties.When Navon and Chen Lin studied the processing of image information in the early visual period,they discovered that the human visual system has the characteristics of prioritizing geometric topological properties,namely the principle of large-scale priority and the theory of topological perception.Although the topological properties are initially the invariant properties of the continuous change of the representational geometry,the essence of the analysis is the embodiment of the invariance in the change.Therefore,since vision has a topological nature,it is widely spread based on existing experience that the entire human cognitive system should also have a topological nature.The content of this article is also to verify the "conjecture of the existence of cognitive structure" proposed by my mentor,that cognitive systems,including five kinds of perception,should be able to integrate the cognitive structure,including topological invariance.This is also an affirmative answer of the fifth basic problem of artificial intelligence,wheather all cognitive systems have an unified structure and lay a foundation for the establishment of the basic theory of Intelligence Science.Secondly,this dissertation takes the topology as the starting point,combines the topological structure of the visual pathway and the auditory pathway,and gradually extends the visual topological properties to the auditory and linguistic systems.Using the idea of topological space in point set topology,the visual and auditory pathway information processing process is demonstrated.Next,we made a neurobiological analysis of the topological mapping processes of the visual,auditory,and linguistic spaces into psychological space.Finally,we combined mental space and extended the topological properties to the human cognitive system to construct a human topology cognitive structure diagram.Then,in this paper,the visual topological properties as the starting point,combined with the calculation model of the early vision,through the normalization model,energy model to simulate the information exchange mechanism of cells in the visual pathway between the same level.This paper proposes to use the Gabor filter to simulate the response process of simple cells as the calculation model of the initial geometrical characteristics of the visual field,and the convolutional neural network to simulate the computational process of the overall geometrical properties of the later visual period.Finally,combining with the three models of Gabor filter,visual information interaction model and convolutional neural network,an early visual topological cognitive model is constructed,and this model is applied to handwritten Chinese character recognition.This paper proposes a way to refine Chinese characters and remove corner points to extract the overall topology of Chinese characters.The final experimental results show that the visual topology cognitive model constructed in this paper is superior to a single convolutional neural network model and has better robustness in handwritten Chinese character recognition.This shows that the topology should be a basic property of the visual system to process information,and it made a preliminary verification of the "conjecture of the existence of cognitive structure" from the visual level,and also began to respond to the fifth basic problem of artificial intelligence.