Evolving artificial neural network ensembles

Using a coordinated group of simple solvers to tackle a complex problem is not an entirely new idea. Its root could be traced back hundreds of years ago when ancient Chinese suggested a team approach to problem solving. For a long time, engineers have used the divide-and-conquer strategy to decompose a complex problem into simpler sub-problems and then solve them by a group of solvers. However, knowing the best way to divide a complex problem into simpler ones relies heavily on the available domain knowledge. It is often a manual process by an experienced engineer. There have been few automatic divide-and-conquer methods reported in the literature. Fortunately, evolutionary computation provides some of the interesting avenues to automatic divide-and-conquer methods. An in-depth study of such methods reveals that there is a deep underlying connection between evolutionary computation and ANN ensembles. Ideas in one area can be usefully transferred into another in producing effective algorithms. For example, using speciation to create and maintain diversity had inspired the development of negative correlation learning for ANN ensembles, and an in-depth study of diversity in ensembles. This paper will review some of the recent work in evolutionary approaches to designing ANN ensembles.     

Reliability assessment of an automated distribution system

Automation can greatly enhance distribution-network reliability by speeding up service restoration and thus significantly reduce customer-outage time. The paper presents an approach to assess quantitatively the adequacy of a particular automated distribution scheme designated as the `low interruption system' (LIS). Owing to the use of a high-speed communication system and line sensors, this automated scheme can reduce drastically the number of interruptions, the service interruption time and also the area affected by the fault. This scheme provides a simple and cost-effective way to automate distribution systems in which the remotely controlled switches speed up isolation of faulted sections and the restoration of healthy sections through alternative routes. The step-by-step calculation procedure is presented using a typical small automated distribution system. The proposed technique is then applied to a larger distribution system to examine the effectiveness of the technique and also to examine the level of reliability improvement achieved by automation     

Learning user preferences of route choice behaviour for adaptive route guidance

As the use of navigation systems becomes more widespread, the demand for advanced functions of navigation systems also increases. In the light of user satisfaction, personalisation of route guidance by incorporating user preferences is one of the most desired features. A user model applied to personalised route guidance is presented. The user model adaptively updates route selection rules when it discovers the predicted choice differs from the actual choice of the driver. This study employs a decision tree learning algorithm, the C4.5 algorithm, which has advantages over other data mining methods in terms of its comprehensible model structure. Simulation experiments with a real-world network were conducted to analyse the applicability of the model to adaptive route guidance and the accuracy of its prediction     

Guest Editors' Introduction: Overview of Sensor Networks

Wireless sensor networks could advance many scientific pursuits while providing a vehicle for enhancing various forms of productivity, including manufacturing, agriculture, construction, and transportation.     

Power management in networked devices

Networks are one of the most significant developments in computing and a hallmark of modern society. However, along with increasing efficiency and productivity, both at home and in the workplace, networks have costs. One cost is the additional energy that electronic devices consume when attached to networks. Power management, a standard feature of modern PCs, was primarily developed to increase battery lifetime in laptop PCs, which historically were not network-connected when using battery power. Today, however, many laptops are connected to a network - typically a Wi-Fi network - as are the majority of desktop computers. Three key drivers of energy use are induced consumption by devices prevented by network connections from entering low-power states, increasing link data rates that inherently consume more energy for the network interfaces, and proliferation of network-connected displays that actively update and display data when no one is present.     

Phase change process with free convection in a circular enclosure: numerical simulations

A numerical study of unsteady natural convection flow during freezing of water in a circular enclosure is presented. Mathematical model for phase change is based on apparent capacity method formulation and the governing equations are discretized on a fixed grid by means of finite element method. Water’s temperature is initially higher than its freezing temperature. Then, the temperature of the enclosure’s boundary is dropped to a temperature lower than freezing temperature. Ice forms at the enclosure boundary while natural convection flow is induced in the liquid region. Calculations have been made for the rate of change of solid fraction and temperature distributions, for conduction and conduction plus convection modes of heat transfer, and density inversion near freezing temperature phenomenon of water is considered. High resolution capturing of solid/liquid moving boundary as well as the details of flow structure is presented. The results indicate that the effect of natural convection is dominant over conduction if the Rayleigh number is higher than 5 × 10⁶ and relatively insignificant if the Rayleigh number is less than 1 × 10⁶.     

Evaluating software reuse alternatives: a model and its application to an industrial case study

We propose a model that enables software developers to systematically evaluate and compare all possible alternative reuse scenarios. The model supports the clear identification of the basic operations involved and associates a cost component with each basic operation in a focused and precise way. The model is a practical tool that assists developers to weigh and evaluate different reuse scenarios, based on accumulated organizational data, and then to decide which option to select in a given situation. The model is currently being used at six different companies for cost-benefit analysis of alternative reuse scenarios; we give a case study that illustrates how it has been used in practice.     

The speed of mouse-click as a measure of anxiety during human-computer interaction

The monitoring of the human-computer interaction process is one of the essential aspects in the evaluation and enhancement of both task and affective outcome of human-computer interaction. However, although objective measures exist for task outcome, most affective measures are subjective. This study represented an investigation into the speed of mouse-click as a possible measure in human-computer interaction, and was based principally on the suggestions that a relationship exists between stress and motor activities involved in the operation of the fingers. Two groups of 30 subjects were exposed to different sets of human-computer interaction conditions, and the speed of mouse-click and state anxiety were examined. No correlation was found between the speed of mouse-click and state anxiety. However, a significant difference was found in the speed of mouse-click between the groups and the different human-computer interaction situations. The implication of these findings and the possible advantages of using the computer mouse to collect data relating to the computer user's covert state during human-computer interaction are discussed.     

Stable social foraging swarms in a noisy environment

Bacteria, bees, and birds often work together in groups to find food. A group of robots can be designed to coordinate their activities to search for and collect objects. Networked cooperative uninhabited autonomous vehicles are being developed for commercial and military applications. Suppose that we refer to all such groups of entities as "social foraging swarms". In order for such multiagent systems to succeed it is often critical that they can both maintain cohesive behaviors and appropriately respond to environmental stimuli (e.g., by optimizing the acquisition of nutrients in foraging for food). In this paper, we characterize swarm cohesiveness as a stability property and use a Lyapunov approach to develop conditions under which local agent actions will lead to cohesive foraging even in the presence of "noise" characterized by uncertainty on sensing other agent's position and velocity, and in sensing nutrients that each agent is foraging for. The results quantify earlier claims that social foraging is in a certain sense superior to individual foraging when noise is present, and provide clear connections between local agent-agent interactions and emergent group behavior. Moreover, the simulations show that very complicated but orderly group behaviors, reminiscent of those seen in biology, emerge in the presence of noise.     

Evaluation of passive microwave snow water equivalent algorithms in the depth hoar-dominated snowpack of the Kuparuk River Watershed, Alaska, USA

This research investigates the utility of passive microwave remote sensing instruments to accurately determine snow water equivalent (SWE) over large spatial extents. Three existing Special Sensor Microwave Imager (SSM/I) snow water equivalent algorithms produced by Chang, Tait and Goodison were evaluated for their ability to determine snow water equivalent in a snowpack containing substantial depth hoar, large faceted snow crystals. The Kuparuk River Watershed (8140 km²) test site on the North Slope of Alaska was chosen for its snowpack containing a think depth hoar layer and long history of ground truth data. A new regional snow water equivalent algorithm was developed to determine if it could produce better results than the existing algorithms in an area known to contain significant depth hoar. The four algorithms were tested to see how well they could determine snow water equivalent: (1) on a per pixel basis, (2) across swath-averaged spatial bands of approximately 850 km², and (3) on a watershed scale. The algorithms were evaluated to see if they captured the annual spatial distribution in snow water equivalent over the watershed. Results show that the algorithms developed by Chang and from this research are generally within 3 cm of the spatially averaged snow water equivalents over the entire watershed. The algorithms produced by Chang, Tait, and in this research were able to predict the basin-wide ground measured snow water equivalent value within a percent error range from −32.4% to 24.4% in the years with a typical snowpack. None of the algorithms produce accurate results on a pixel-by-pixel scale, with errors ranging from −26% to 308%.