Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
A Comprehensive Survey on Higher Order Neural Networks and Evolutionary Optimization Learning Algorithms in Financial Time Series Forecasting
https://orcid.org/0000-0003-0804-1264
Abstract The financial market volatility has been a focus of study for experts over past decades. While stockbrokers and investors expect reliable projections of future stock indices, it instead displays unpredictable, complicated, and nonlinear reactions which paves path towards designing accurate prediction mechanisms for approximate stock market behaviour. Many computational intelligence methods have been used in the field of economic forecasting alongside more conventional methods. Yet, a trustworthy forecasting depends heavily on the correct selection of an efficient model and optimum set of model hyperparameters. In this regard, artificial neural networks (ANNs) specifically higher-order neural networks (HONNs) have shown notable performance in modelling the chaotic behaviour of financial data and predicting future data. Improved precision and nonlinear decision boundaries are two benefits of using HONN. They can store more data, map features beautifully with weights that are configurable on a single layer, learn quickly, and even tackle complex real-world situations. Besides, evolutionary optimization algorithms (EOA) are widely employed for adjusting HONN parameters thus, enhancing their generalization capability. HONNs are trained with several flavours of evolutionary learning algorithms to discover optimal solutions for various real-world complex engineering problems. The purpose of this article is to explore HONN and EOA based financial forecasts from the last two decades and see how well these forecasts performed in terms of making accurate and reliable financial predictions. We conducted a rigorous survey on state-of-the-art HONN and EOA based methods used for financial forecasting collected from reliable and reputed sources, analysed their applicability and performability, identified their strength and limitations, and suggested few constructive criticisms. Findings of this study may assist researchers of this field in adopting such forecasting methods appropriately.
A Comprehensive Survey on Higher Order Neural Networks and Evolutionary Optimization Learning Algorithms in Financial Time Series Forecasting
https://orcid.org/0000-0003-0804-1264
Abstract The financial market volatility has been a focus of study for experts over past decades. While stockbrokers and investors expect reliable projections of future stock indices, it instead displays unpredictable, complicated, and nonlinear reactions which paves path towards designing accurate prediction mechanisms for approximate stock market behaviour. Many computational intelligence methods have been used in the field of economic forecasting alongside more conventional methods. Yet, a trustworthy forecasting depends heavily on the correct selection of an efficient model and optimum set of model hyperparameters. In this regard, artificial neural networks (ANNs) specifically higher-order neural networks (HONNs) have shown notable performance in modelling the chaotic behaviour of financial data and predicting future data. Improved precision and nonlinear decision boundaries are two benefits of using HONN. They can store more data, map features beautifully with weights that are configurable on a single layer, learn quickly, and even tackle complex real-world situations. Besides, evolutionary optimization algorithms (EOA) are widely employed for adjusting HONN parameters thus, enhancing their generalization capability. HONNs are trained with several flavours of evolutionary learning algorithms to discover optimal solutions for various real-world complex engineering problems. The purpose of this article is to explore HONN and EOA based financial forecasts from the last two decades and see how well these forecasts performed in terms of making accurate and reliable financial predictions. We conducted a rigorous survey on state-of-the-art HONN and EOA based methods used for financial forecasting collected from reliable and reputed sources, analysed their applicability and performability, identified their strength and limitations, and suggested few constructive criticisms. Findings of this study may assist researchers of this field in adopting such forecasting methods appropriately.
A Comprehensive Survey on Higher Order Neural Networks and Evolutionary Optimization Learning Algorithms in Financial Time Series Forecasting
https://orcid.org/0000-0003-0804-1264
Abstract The financial market volatility has been a focus of study for experts over past decades. While stockbrokers and investors expect reliable projections of future stock indices, it instead displays unpredictable, complicated, and nonlinear reactions which paves path towards designing accurate prediction mechanisms for approximate stock market behaviour. Many computational intelligence methods have been used in the field of economic forecasting alongside more conventional methods. Yet, a trustworthy forecasting depends heavily on the correct selection of an efficient model and optimum set of model hyperparameters. In this regard, artificial neural networks (ANNs) specifically higher-order neural networks (HONNs) have shown notable performance in modelling the chaotic behaviour of financial data and predicting future data. Improved precision and nonlinear decision boundaries are two benefits of using HONN. They can store more data, map features beautifully with weights that are configurable on a single layer, learn quickly, and even tackle complex real-world situations. Besides, evolutionary optimization algorithms (EOA) are widely employed for adjusting HONN parameters thus, enhancing their generalization capability. HONNs are trained with several flavours of evolutionary learning algorithms to discover optimal solutions for various real-world complex engineering problems. The purpose of this article is to explore HONN and EOA based financial forecasts from the last two decades and see how well these forecasts performed in terms of making accurate and reliable financial predictions. We conducted a rigorous survey on state-of-the-art HONN and EOA based methods used for financial forecasting collected from reliable and reputed sources, analysed their applicability and performability, identified their strength and limitations, and suggested few constructive criticisms. Findings of this study may assist researchers of this field in adopting such forecasting methods appropriately.
A Comprehensive Survey on Higher Order Neural Networks and Evolutionary Optimization Learning Algorithms in Financial Time Series Forecasting
Behera, Sudersan (Autor:in) / Nayak, Sarat Chandra (Autor:in) / Kumar, A. V. S. Pavan (Autor:in)
2023
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
River-Flow Forecasting Using Higher-Order Neural Networks
| Online Contents | 2012
River-Flow Forecasting Using Higher-Order Neural Networks
| British Library Online Contents | 2012
Artificial Neural Networks for Forecasting of Fuzzy Time Series
| Online Contents | 2010