Results

**1 - 4**of**4**### Orthogonal-Least-Squares Forward Selection for Parsimonious Modelling from Data

, 2009

"... The objective of modelling from data is not that the model simply fits the training data well. Rather, the goodness of a model is characterized by its generalization capability, interpretability and ease for knowledge extraction. All these desired properties depend crucially on the ability to constr ..."

Abstract
- Add to MetaCart

(Show Context)
The objective of modelling from data is not that the model simply fits the training data well. Rather, the goodness of a model is characterized by its generalization capability, interpretability and ease for knowledge extraction. All these desired properties depend crucially on the ability to construct appropriate parsimonious models by the modelling process, and a basic principle in practical nonlinear data modelling is the parsimonious principle of ensuring the smallest possible model that explains the training data. There exists a vast amount of works in the area of sparse modelling, and a widely adopted approach is based on the linear-in-the-parameters data modelling that include the radial basis function network, the neurofuzzy network and all the sparse kernel modelling techniques. A well tested strategy for parsimonious modelling from data is the orthogonal least squares (OLS) algorithm for forward selection modelling, which is capable of constructing sparse models that generalise well. This contribution continues this theme and provides a unified framework for sparse modelling from data that includes regression and classification, which belong to supervised learning, and probability density function estimation, which is an unsupervised learning problem. The OLS forward selection method based on the leave-one-out test criteria is presented within this unified data-modelling framework. Examples from regression, classification and density estimation applications are used to illustrate the effectiveness of this generic parsimonious modelling approach from data.

### APPLICATION OF COMPUTATIONAL INTELLIGENCE TO TARGET TRACKING

"... In the oceanic context, the aim of Target Motion Analysis (TMA) is to estimate the state, i.e. location, bearing and velocity, of a sound-emitting object. These estimates are based on a series of passive measures of both the angle and the distance between an observer and the source of sound, which i ..."

Abstract
- Add to MetaCart

(Show Context)
In the oceanic context, the aim of Target Motion Analysis (TMA) is to estimate the state, i.e. location, bearing and velocity, of a sound-emitting object. These estimates are based on a series of passive measures of both the angle and the distance between an observer and the source of sound, which is called the target. These measurements are corrupted by noise and false readings, which are perceived as outliers. Usually, sequences of measurements are taken and statistical methods, like the Least Squares method or the Annealing M-Estimator, are applied to estimate the target's state by minimising the residual in range and bearing for a series of measurements. In this project, an ACO-Estimator, a novel hybrid optimisation algorithm based on Computational Intelligence, has been developed and applied to the TMA problem and its effectiveness was compared with standard estimators. It was shown that the new algorithm outperforms conventional estimators by successfully removing outliers from the measurements.

### Federal Funds Rate Prediction: A Comparison Between the Robust RBF Neural Network and Economic Models*

"... Neural network forecasting models have been widely used in the analyses of finan-cial time series during the last decade. This paper attempts to fill this gap in the literature by examining a variety of univariate and multivariate, linear, nonlinear Economics em-pirical modes and neural network. In ..."

Abstract
- Add to MetaCart

(Show Context)
Neural network forecasting models have been widely used in the analyses of finan-cial time series during the last decade. This paper attempts to fill this gap in the literature by examining a variety of univariate and multivariate, linear, nonlinear Economics em-pirical modes and neural network. In this paper, we construct an M-estimator based RBF (MRBF) neural network with growing and pruning techniques. Then we compare the forecasting performances of MRBF with six other time-series forecasting models for daily U.S. effective federal funds rate. The results show that the proposed MRBF net-work can produce the lowest root mean square errors in one-day-ahead forecasting for the federal funds rate. It implies that MRBF can be one good method for the predictions of some financial time series data.

### Model Structure Selection Using an Integrated Forward Orthogonal Search Algorithm Assisted by Squared Correlation and Mutual Information

"... promoting access to White Rose research papers ..."

(Show Context)