[4K3-GS-1] Fundamental AI, theory: cognition / perception

[4K3-GS-1-02] Dynamic properties of theory of indeterminate natural transformation

〇Takaomi Yokosuka¹, Miho Fuyama², Hayato Saigo³, Tatsuji Takahashi^1,4 (1. Tokyo Denki University, 2. Waseda University, 3. Nagahama Institute of Bio-Science and Technology, 4. RIKEN AIP)

Keywords:category theory, complex network, association, analogy, metaphor comprehension

Metaphors play important roles in conceptual representation and language comprehension, as a way of cognitive processes such as "compare X (unknown object) to Y (known object) ".
The authors propose the theory of indeterminate natural transformation (TINT) as a model of metaphor comprehension. The underlying idea of the model is that the meaning of an image can be described as an indeterminate category that introduces a stochastic process into category theory. Previous studies have only proposed tentative rules for the process of excitation and relaxation in the indeterminate category.
In this study, simulations were implemented and analyzed the performances of the results , to investigate the nature of the network that results from the operations of the TINT that changes the indeterminate category.
The results show that TINT increases the number of objects while maintaining a constant network structure, and generates a scale-free network from a random network.