The evolving Mamdani-Takagi-Sugeno neural-fuzzy inference system and its applications in the financial domain

• Main Author: Ho, Stanley Weng Luen.
• Other Authors: Quek Hiok Chai
• Format: Final Year Project
• Language: English
• Published: 2010
• Subjects: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence
• Online Access: http://hdl.handle.net/10356/40209
• Institution: Nanyang Technological University

This thesis presents a novel neural-fuzzy network architecture named the evolving Mamdani-Takagi-Sugeno neural fuzzy inference system (eMTSFIS) that seeks to address two deficiencies faced by neural-fuzzy systems. Firstly, the dynamic nature of real-world problems demands that neural-fuzzy systems be able to adapt their parameters and evolve their rule-bases to address the time-varying characteristics of their operating environments. Secondly, in practice, having good fuzzy rule-base interpretability and high modeling accuracy are contradictory requirements and one usually prevails over the other based on the modeling objective and fuzzy rule structure employed. The eMTSFIS model is able to achieve life-long learning as it evolves and adapts its knowledge to the dynamics of the underlying environment. This effectively addresses the stability-plasticity dilemma. The eMTSFIS model combines Mamdani and T-S fuzzy modeling approaches, coupled with a localized parameter learning approach, to achieve both improved interpretability and accuracy. Experimental results from three benchmark applications demonstrate the learning robustness and modeling versatility of the eMTSFIS model. The results are encouraging. This project also presents its applications in the financial domain. First, an intraday high frequency financial trading system with the eMTSFIS predictive model is proposed. This trading system incorporates the learning robustness and modeling capability of the eMTSFIS predictive model, as it is able to evolve its system structure to cope with the dynamic nature of intraday high frequency trading. In addition, stock Liquidity Reversal was incorporated into the trading system. Collectively, the improvised trading system was shown to be a suitable tool that can be used in the intraday trading environment to compliment existing technical indicators, such as MACD and RSI. More importantly, the improvised trading system was able to outperform all other trading systems that use well-established technical indicators, in times of both bearish market sentiment and bullish market sentiment. Second, an interday financial trading system with the eMTSFIS predictive model is proposed. This trading system was experimented on various financial vehicles, namely major market indices, foreign exchange rates and a major stock counter. The trading system was shown to make wiser trading decisions and to generate higher returns for the investor.