Results 1 
7 of
7
M.N.Vrahatis, Financial forecasting through unsupervised clustering and evolutionary trained neural networks
 in: Congress on Evolutionary Computation
, 2003
"... In this paper, we review our work on a time series forecasting methodology based on the combination of unsupervised clustering and artificial neural networks. To address noise and nonstationarity, a common approach is to combine a method for the partitioning of the input space into a number of subs ..."
Abstract

Cited by 15 (8 self)
 Add to MetaCart
In this paper, we review our work on a time series forecasting methodology based on the combination of unsupervised clustering and artificial neural networks. To address noise and nonstationarity, a common approach is to combine a method for the partitioning of the input space into a number of subspaces with a local approximation scheme for each subspace. Unsupervised clustering algorithms have the desirable property of deciding on the number of partitions required to accurately segment the input space during the clustering process, thus relieving the user from making this ad hoc choice. Artificial neural networks, on the other hand, are powerful computational models that have proved their capabilities on numerous hard realworld problems. The time series that we consider are all daily spot foreign exchange rates of major currencies. The experimental results reported suggest that predictability varies across different regions of the input space, irrespective of clustering algorithm. In all cases, there are regions that are associated with a particularly high forecasting performance. Evaluating the performance of the proposed methodology with respect to its profit generating capability indicates that it compares favorably with that of two other established approaches. Moving from the task of onestepahead to multiplestepahead prediction, performance deteriorates rapidly.
Time Series Prediction with the SelfOrganizing Map: A Review
"... Summary. We provide a comprehensive and updated survey on applications of Kohonen’s selforganizing map (SOM) to time series prediction (TSP). The main goal of the paper is to show that, despite being originally designed as an unsupervised learning algorithm, the SOM is flexible enough to give rise ..."
Abstract

Cited by 2 (0 self)
 Add to MetaCart
Summary. We provide a comprehensive and updated survey on applications of Kohonen’s selforganizing map (SOM) to time series prediction (TSP). The main goal of the paper is to show that, despite being originally designed as an unsupervised learning algorithm, the SOM is flexible enough to give rise to a number of efficient supervised neural architectures devoted to TSP tasks. For each SOMbased architecture to be presented, we report its algorithm implementation in detail. Similarities and differences of such SOMbased TSP models with respect to standard linear and nonlinear TSP techniques are also highlighted. We conclude the paper with indications of possible directions for further research on this field. Key words: Selforganizing map, time series prediction, local linear mappings, vectorquantized temporal associative memory, radial basis functions. 1
Vrahatis, Computational intelligence methods for financial forecasting
 Proceedings of the International Conference of Computational Methods in Sciences and Engineering (ICCMSE 2005
"... Abstract: Forecasting the short run behavior of foreign exchange rates is a challenging problem that has attracted considerable attention. High frequency financial data are typically characterized by noise and non–stationarity. In this work we investigate the profitability of a forecasting methodolo ..."
Abstract

Cited by 1 (1 self)
 Add to MetaCart
Abstract: Forecasting the short run behavior of foreign exchange rates is a challenging problem that has attracted considerable attention. High frequency financial data are typically characterized by noise and non–stationarity. In this work we investigate the profitability of a forecasting methodology based on unsupervised clustering and feedforward neural networks and compare its performance with that of a single feedforward neural network and nearest neighbor regression. The experimental results indicate that the proposed combination of the two methodologies achieves a higher profit. Keywords:
Researcher Researcher Researcher
"... Financial forecasting has been challenging problem due to its high nonlinearity and high volatility. An Artificial Neural Network (ANN) can model flexible linear or nonlinear relations hip among variables. ANN can be configured to produce desired set of output based on set of given input. In this ..."
Abstract
 Add to MetaCart
Financial forecasting has been challenging problem due to its high nonlinearity and high volatility. An Artificial Neural Network (ANN) can model flexible linear or nonlinear relations hip among variables. ANN can be configured to produce desired set of output based on set of given input. In this paper we attempt at analyzing the usefulness of artificial neural network for forecasting financial data series with use of different algorithms such as backpropagation, radial basis function etc. With their ability of adapting nonlinear and chaotic patterns, ANN is the current technique being used which offers the ability of predicting financial data more accurately. "A xy1 network topology is adopted because of x input variables in which variable y was determined by the number of hidden neurons during network selection with single output. " Both x and y were changed.
unknown title
, 2006
"... www.elsevier.com/locate/mcm Generalizing the kWindows clustering algorithm in metric spaces ..."
Abstract
 Add to MetaCart
www.elsevier.com/locate/mcm Generalizing the kWindows clustering algorithm in metric spaces
An applied methodology for the prediction of time series ’ local optima
"... Time series prediction is directly connected with the major problem of portfolio optimization; its solution is detected in numerous research and in many statistical methodologies which have implied towards the investigation of the best set of assets. In finance, the time series theory is mainly appl ..."
Abstract
 Add to MetaCart
Time series prediction is directly connected with the major problem of portfolio optimization; its solution is detected in numerous research and in many statistical methodologies which have implied towards the investigation of the best set of assets. In finance, the time series theory is mainly applied for the prediction of the stock market prices or in applications regarding the currency levels. Recently, have proposed gradient unconstrained optimization algorithms that are being used in the process of the Lipschitz constant estimation towards the approximation of the objective functions optima. More detailed, the estimation of the Lipschitz constant is calculated on sequenced points that come from the repetitive process of optima finding, and their function values, as well. Furthermore, it is proved that the use of this step size, given from the Lipschitz constant estimation, leads to local optimum point. This paper attempts to forecast a time series future optima by applying the Steepest Descent with the Adaptive Step size (SDAS) algorithm. Thus, n − 1 past and known points from the time series are chosen in such way that all necessary conditions are applied; furthermore these points represent sequence points of a repetitive process that leads to local optima of the objective function. The proposed methodology was tested on the daily closing prices of the Athens ’ Stock Market. The results obtained provide clues that the proposed methodology predicts the local maxima and minima in a rather successive rate. What, however, should be furtherly investigated is the degree that each characteristic of the sample and their occasional fluctuations may affect the results ’ accuracy.
Proceedings of the International Multiconference on
"... Estimating time series future optima using a steepest descent methodology as a backtracker ..."
Abstract
 Add to MetaCart
Estimating time series future optima using a steepest descent methodology as a backtracker