Use of a probabilistic neural network to reduce costs of selecting construction rock

Rocks used as construction aggregate in temperate climates deteriorate to differing degrees because of repeated freezing and thawing. The magnitude of the deterioration depends on the rock's properties. Aggregate, including crushed carbonate rock, is required to have minimum geotechnical qualities before it can be used in asphalt and concrete. In order to reduce chances of premature and expensive repairs, extensive freeze-thaw tests are conducted on potential construction rocks. These tests typically involve 300 freeze-thaw cycles and can take four to five months to complete. Less time consuming tests that (1) predict durability as well as the extended freeze-thaw test or that (2) reduce the number of rocks subject to the extended test, could save considerable amounts of money. Here we use a probabilistic neural network to try and predict durability as determined by the freeze-thaw test using four rock properties measured on 843 limestone samples from the Kansas Department of Transportation. Modified freeze-thaw tests and less time consuming specific gravity (dry), specific gravity (saturated), and modified absorption tests were conducted on each sample. Durability factors of 95 or more as determined from the extensive freeze-thaw tests are viewed as acceptable—rocks with values below 95 are rejected. If only the modified freeze-thaw test is used to predict which rocks are acceptable, about 45% are misclassified. When 421 randomly selected samples and all four standardized and scaled variables were used to train aprobabilistic neural network, the rate of misclassification of 422 independent validation samples dropped to 28%. The network was trained so that each class (group) and each variable had its own coefficient (sigma). In an attempt to reduce errors further, an additional class was added to the training data to predict durability values greater than 84 and less than 98, resulting in only 11% of the samples misclassified. About 43% of the test data was classed by the neural net into the middle group—these rocks should be subject to full freeze-thaw tests. Thus, use of the probabilistic neural network would meanthat the extended test would only need be applied to 43% of the samples, and 11% of the rocks classed as acceptable would fail early.

Events in Indonesia: exploring the limits to formal tourism trends forecasting methods in complex crisis situations

The desire to know the future is as old as humanity. For the tourism industry the demand for accurate foretelling of the future course of events is a task that consumes considerable energy and is of great significance to investors. This paper examines the issue of forecasting by comparing forecasts of inbound tourism made prior to the political and economic crises that engulfed Indonesia from 1997 onwards with actual arrival figures. The paper finds that current methods of forecasting are not able to cope with unexpected crises and other disasters and that alternative methods need to be examined including scenarios, political risk and application of chaos theory. The paper outlines a framework for classifying shocks according to a scale of severity, probability, type of event, level of certainty and suggested forecasting tools for each scale of shock. (C) 2003 Elsevier Science Ltd. All rights reserved.

Forecasting with serially correlated regression models

In this article we investigate the asymptotic and finite-sample properties of predictors of regression models with autocorrelated errors. We prove new theorems associated with the predictive efficiency of generalized least squares (GLS) and incorrectly structured GLS predictors. We also establish the form associated with their predictive mean squared errors as well as the magnitude of these errors relative to each other and to those generated from the ordinary least squares (OLS) predictor. A large simulation study is used to evaluate the finite-sample performance of forecasts generated from models using different corrections for the serial correlation.

Forecasting stock returns using model-selection criteria

This paper examines the economic significance of return predictability in Australian equities. In light of considerable model uncertainty, formal model-selection criteria are used to choose a specification for the predictive model. A portfolio-switching strategy is implemented according to model predictions. Relative to a buy-and-hold market investment, the returns to the portfolio-switching strategy are impressive under several model-selection criteria, even after accounting for transaction costs. However, as these findings are not robust across other model-selection criteria examined, it is difficult to conclude that the degree of return predictability is economically significant.

A simple regional-scale model for forecasting sorghum yield across North-Eastern Australia

Sorghum is the main dryland summer crop in NE Australia and a number of agricultural businesses would benefit from an ability to forecast production likelihood at regional scale. In this study we sought to develop a simple agro-climatic modelling approach for predicting shire (statistical local area) sorghum yield. Actual shire yield data, available for the period 1983-1997 from the Australian Bureau of Statistics, were used to train the model. Shire yield was related to a water stress index (SI) that was derived from the agro-climatic model. The model involved a simple fallow and crop water balance that was driven by climate data available at recording stations within each shire. Parameters defining the soil water holding capacity, maximum number of sowings (MXNS) in any year, planting rainfall requirement, and critical period for stress during the crop cycle were optimised as part of the model fitting procedure. Cross-validated correlations (CVR) ranged from 0.5 to 0.9 at shire scale. When aggregated to regional and national scales, 78-84% of the annual variation in sorghum yield was explained. The model was used to examine trends in sorghum productivity and the approach to using it in an operational forecasting system was outlined. (c) 2005 Elsevier B.V. All rights reserved.

A probabilistic switch model of information flow in social networks: Application for virtual circuits to organizational design

Network building and exchange of information by people within networks is crucial to the innovation process. Contrary to older models, in social networks the flow of information is noncontinuous and nonlinear. There are critical barriers to information flow that operate in a problematic manner. New models and new analytic tools are needed for these systems. This paper introduces the concept of virtual circuits and draws on recent concepts of network modelling and design to introduce a probabilistic switch theory that can be described using matrices. It can be used to model multistep information flow between people within organisational networks, to provide formal definitions of efficient and balanced networks and to describe distortion of information as it passes along human communication channels. The concept of multi-dimensional information space arises naturally from the use of matrices. The theory and the use of serial diagonal matrices have applications to organisational design and to the modelling of other systems. It is hypothesised that opinion leaders or creative individuals are more likely to emerge at information-rich nodes in networks. A mathematical definition of such nodes is developed and it does not invariably correspond with centrality as defined by early work on networks.

Forecasting Helicoverpa populations in Australia: A comparison of regression based models and a bio-climatic based modeling approach

Long-term forecasts of pest pressure are central to the effective management of many agricultural insect pests. In the eastern cropping regions of Australia, serious infestations of Helicoverpa punctigera (Wallengren) and H. armigera (Hübner)(Lepidoptera: Noctuidae) are experienced annually. Regression analyses of a long series of light-trap catches of adult moths were used to describe the seasonal dynamics of both species. The size of the spring generation in eastern cropping zones could be related to rainfall in putative source areas in inland Australia. Subsequent generations could be related to the abundance of various crops in agricultural areas, rainfall and the magnitude of the spring population peak. As rainfall figured prominently as a predictor variable, and can itself be predicted using the Southern Oscillation Index (SOI), trap catches were also related to this variable. The geographic distribution of each species was modelled in relation to climate and CLIMEX was used to predict temporal variation in abundance at given putative source sites in inland Australia using historical meteorological data. These predictions were then correlated with subsequent pest abundance data in a major cropping region. The regression-based and bioclimatic-based approaches to predicting pest abundance are compared and their utility in predicting and interpreting pest dynamics are discussed.

Population-based continuous optimization, probabilistic modelling and mean shift

Evolutionary algorithms perform optimization using a population of sample solution points. An interesting development has been to view population-based optimization as the process of evolving an explicit, probabilistic model of the search space. This paper investigates a formal basis for continuous, population-based optimization in terms of a stochastic gradient descent on the Kullback-Leibler divergence between the model probability density and the objective function, represented as an unknown density of assumed form. This leads to an update rule that is related and compared with previous theoretical work, a continuous version of the population-based incremental learning algorithm, and the generalized mean shift clustering framework. Experimental results are presented that demonstrate the dynamics of the new algorithm on a set of simple test problems.

Vector Error-Correction Models in a consumer packaged goods category forecasting decision support system

A framework for developing marketing category management decision support systems (DSS) based upon the Bayesian Vector Autoregressive (BVAR) model is extended. Since the BVAR model is vulnerable to permanent and temporary shifts in purchasing patterns over time, a form that can correct for the shifts and still provide the other advantages of the BVAR is a Bayesian Vector Error-Correction Model (BVECM). We present the mechanics of extending the DSS to move from a BVAR model to the BVECM model for the category management problem. Several additional iterative steps are required in the DSS to allow the decision maker to arrive at the best forecast possible. The revised marketing DSS framework and model fitting procedures are described. Validation is conducted on a sample problem.

Power system sensitivity analysis for probabilistic small signal stability assessment in a deregulated environment

Deregulations and market practices in power industry have brought great challenges to the system planning area. In particular, they introduce a variety of uncertainties to system planning. New techniques are required to cope with such uncertainties. As a promising approach, probabilistic methods are attracting more and more attentions by system planners. In small signal stability analysis, generation control parameters play an important role in determining the stability margin. The objective of this paper is to investigate power system state matrix sensitivity characteristics with respect to system parameter uncertainties with analytical and numerical approaches and to identify those parameters have great impact on system eigenvalues, therefore, the system stability properties. Those identified parameter variations need to be investigated with priority. The results can be used to help Regional Transmission Organizations (RTOs) and Independent System Operators (ISOs) perform planning studies under the open access environment.

Matching of catalogues by probabilistic pattern classification

We consider the statistical problem of catalogue matching from a machine learning perspective with the goal of producing probabilistic outputs, and using all available information. A framework is provided that unifies two existing approaches to producing probabilistic outputs in the literature, one based on combining distribution estimates and the other based on combining probabilistic classifiers. We apply both of these to the problem of matching the HI Parkes All Sky Survey radio catalogue with large positional uncertainties to the much denser SuperCOSMOS catalogue with much smaller positional uncertainties. We demonstrate the utility of probabilistic outputs by a controllable completeness and efficiency trade-off and by identifying objects that have high probability of being rare. Finally, possible biasing effects in the output of these classifiers are also highlighted and discussed.

Prediction of protein continuum secondary structure with probabilistic models based on NMR solved structures

Background: The structure of proteins may change as a result of the inherent flexibility of some protein regions. We develop and explore probabilistic machine learning methods for predicting a continuum secondary structure, i.e. assigning probabilities to the conformational states of a residue. We train our methods using data derived from high-quality NMR models. Results: Several probabilistic models not only successfully estimate the continuum secondary structure, but also provide a categorical output on par with models directly trained on categorical data. Importantly, models trained on the continuum secondary structure are also better than their categorical counterparts at identifying the conformational state for structurally ambivalent residues. Conclusion: Cascaded probabilistic neural networks trained on the continuum secondary structure exhibit better accuracy in structurally ambivalent regions of proteins, while sustaining an overall classification accuracy on par with standard, categorical prediction methods.

Probabilistic world modelling for distributed team planning

This paper describes an application of decoupled probabilistic world modeling to achieve team planning. The research is based on the principle that tbe action selection mechanism of a member in a robot team cm select am effective action if a global world model is available to all team members. In the real world, the sensors are imprecise, and are individual to each robot, hence providing each robot a partial and unique view about the environment. We address this problem by creating a probabilistic global view on each agent by combining the perceptual information from each robot. This probsbilistie view forms the basis for selecting actions to achieve the team goal in a dynamic environment. Experiments have been carried ont to investigate the effectiveness of this principle using custom-built robots for real world performance, in addition, to extensive simulation results. The results show an improvement in team effectiveness when using probabilistic world modeling based on perception sharing for team planning.