科研成果详情

In this paper, the author proposes an innovative Chaotic Interval Type-2 Fuzzy Neuro-oscillatory Network (CIT2-FNON) for worldwide financial prediction. Inspired by the author’s original work on Lee-oscillator—a chaotic discrete-time neural oscillator with profound transient-chaotic property, CIT2-FNON provides: (1) effective modeling of Interval Type-2 Fuzzy Logic with Chaotic Transient-Fuzzy Membership Function (CTFMF); and (2) time-series recurrent neural network training and prediction with Chaotic Bifurcation Transfer Function (CBTF). Different from the contemporary research on Type-2 Fuzzy Logic Systems (T2FLS) which mainly focus on the R&D of the Interval Type-2 Fuzzy Logic (IT2FL)—a simplified version of T2FLS due to its computational complexity, the main innovation of this paper include: (1) the Chaotic Type-2 Transient-Fuzzy Logic (CT2TFL) proposed in this paper provides a truly T2FLS with remarkable chaotic transient-fuzzy property to resolve the computational complexity problem; and (2) different from contemporary fuzzy-neuro systems which focus on the integration of fuzzy logic and neural networks as separated functional modules, the CIT2-FNON introduced in this paper is constructed by Lee-oscillators which serves as “transient-fuzzy input neurons” of the recurrent network and effectively converts it into CT2TFL system. In other words, the chaotic transient-fuzzification process is actually part of the neural model of the CIT2-FNON. From the implementation perspective, CIT2-FNON is integrated with 2048-trading day time-series financial data and Top-10 major financial signals as financial fuzzy signals (FFS) for the real-time prediction of 129 worldwide financial products.