GYAN: A methodology for rule extraction from artificial neural networks

Artificial neural network (ANN) learning methods provide a robust and non-linear approach to approximating the target function for many classification, regression and clustering problems. ANNs have demonstrated good predictive performance in a wide variety of practical problems. However, there are strong arguments as to why ANNs are not sufficient for the general representation of knowledge. The arguments are the poor comprehensibility of the learned ANN, and the inability to represent explanation structures. The overall objective of this thesis is to address these issues by: (1) explanation of the decision process in ANNs in the form of symbolic rules (predicate rules with variables); and (2) provision of explanatory capability by mapping the general conceptual knowledge that is learned by the neural networks into a knowledge base to be used in a rule-based reasoning system. A multi-stage methodology GYAN is developed and evaluated for the task of extracting knowledge from the trained ANNs. The extracted knowledge is represented in the form of restricted first-order logic rules, and subsequently allows user interaction by interfacing with a knowledge based reasoner. The performance of GYAN is demonstrated using a number of real world and artificial data sets. The empirical results demonstrate that: (1) an equivalent symbolic interpretation is derived describing the overall behaviour of the ANN with high accuracy and fidelity, and (2) a concise explanation is given (in terms of rules, facts and predicates activated in a reasoning episode) as to why a particular instance is being classified into a certain category.

The concept of vulnerability and the protection of human subjects of research

This article provides an overview of the concept of vulnerability through the lens of the U.S. federal regulations for the protection of human subjects of research. General issues that emerge for nurse researchers working with regulated vulnerable populations are identified. Points of current controversy in the application of the regulations and current discourse about vulnerable groups are highlighted. Suggestions for negotiating the tension between federally regulated human subject requirements and the realities of research with vulnerable subjects are given. The limitations of the designation of vulnerable as a protection for human subjects will also be discussed.

Soil carbon saturation: Linking concept and measurable carbon pools

The soil C saturation concept suggests a limit to whole soil organic carbon (SOC) accumulation determined by inherent physicochemical characteristics of four soil C pools: unprotected, physically protected, chemically protected, and biochemically protected. Previous attempts to quantify soil C sequestration capacity have focused primarily on silt and clay protection and largely ignored the effects of soil structural protection and biochemical protection. We assessed two contrasting models of SOC accumulation, one with no saturation limit (i.e., linear first-order model) and one with an explicit soil C saturation limit (i.e., C saturation model). We isolated soil fractions corresponding to the C pools (i.e., free particulate organic matter POM], microaggregate-associated C, silt- and clay-associated C, and non-hydrolyzable C) from eight long-term agroecosystern experiments across the United States and Canada. Due to the composite nature of the physically protected C pool, we firactioned it into mineral- vs. POM-associated C. Within each site, the number of fractions fitting the C saturation model was directly related to maximum SOC content, suggesting that a broad range in SOC content is necessary to evaluate fraction C saturation. The two sites with the greatest SOC range showed C saturation behavior in the chemically, biochemically, and some mineral-associated fractions of the physically protected pool. The unprotected pool and the aggregate-protected POM showed linear, nonsaturating behavior. Evidence of C saturation of chemically and biochemically protected SOC pools was observed at sites far from their theoretical C saturation level, while saturation of aggregate-protected fractions occurred in soils closer to their C saturation level.

Soil carbon saturation : concept, evidence and evaluation

Current estimates of soil C storage potential are based on models or factors that assume linearity between C input levels and C stocks at steady-state, implying that SOC stocks could increase without limit as C input levels increase. However, some soils show little or no increase in steady-state SOC stock with increasing C input levels suggesting that SOC can become saturated with respect to C input. We used long-term field experiment data to assess alternative hypotheses of soil carbon storage by three simple models: a linear model (no saturation), a one-pool whole-soil C saturation model, and a two-pool mixed model with C saturation of a single C pool, but not the whole soil. The one-pool C saturation model best fit the combined data from 14 sites, four individual sites were best-fit with the linear model, and no sites were best fit by the mixed model. These results indicate that existing agricultural field experiments generally have too small a range in C input levels to show saturation behavior, and verify the accepted linear relationship between soil C and C input used to model SOM dynamics. However, all sites combined and the site with the widest range in C input levels were best fit with the C-saturation model. Nevertheless, the same site produced distinct effective stabilization capacity curves rather than an absolute C saturation level. We conclude that the saturation of soil C does occur and therefore the greatest efficiency in soil C sequestration will be in soils further from C saturation.

A new pulsed power supply topology based on positive buck-boost converters concept

Improving efficiency and flexibility in pulsed power supply technologies is the most substantial concern of pulsed power systems specifically with regard to plasma generation. Recently, the improvement of pulsed power supply has become of greater concern due to the extension of pulsed power applications to environmental and industrial areas. With this respect, a current source based topology is proposed in this paper as a pulsed power supply which gives the possibility of power flow control during load supplying mode. The main contribution in this configuration is utilization of low-medium voltage semiconductor switches for high voltage generation. A number of switch-diode-capacitor units are designated at the output of topology to exchange the current source energy into voltage form and generate a pulsed power with sufficient voltage magnitude and stress. Simulations carried out in Matlab/SIMULINK platform as well as experimental tests on a prototype setup have verified the capability of this topology in performing desired duties. Being efficient and flexible are the main advantages of this topology.

Applying the open innovation system concept to infrastructure projects

Purpose: The goal of this conceptual paper is to provide tools to help maximise the value delivered by infrastructure projects, by developing methods to increase adoption of innovative products during construction. Methods: The role of knowledge flows in determining innovation adoption rates is conceptually examined. A promising new approach is developed. Open innovation system theory is extended, by reviewing the role of three frameworks: (1) knowledge intermediaries, (2) absorptive capacity and (3) governance arrangements. Originality: We develop a novel open innovation system model to guide further research in the area of adoption of innovation on infrastructure projects. The open innovation system model currently lacks definition of core concepts, especially with regard to the impact of different degrees and types of openness. The three frameworks address this issue and add substance to the open innovation system model, addressing widespread criticism that it is underdeveloped. The novelty of our model is in the combination of the three frameworks to explore the system. These frameworks promise new insights into system dynamics and facilitate the development of new methods to optimise the diffusion of innovation. Practical Implications: The framework will help to reveal gaps in knowledge flows that impede the uptake of innovations. In the past, identifying these gaps has been difficult given the lack of nuance in existing theory. The knowledge maps proposed will enable informed policy advice to effectively harness the power of knowledge networks, increase innovation diffusion and improve the performance of infrastructure projects. The models developed in this paper will be used in planned empirical research into innovation on large scale infrastructure projects in the Australian built environment.

Concept building

In this world of continuous change, there’s probably one certainty: more change lies ahead. Our students will encounter challenges and opportunities that we can’t even imagine. How do we prepare our students as future citizens for the challenges of the 21st century? One of the most influential public intellectuals of our time, Howard Gardner, suggests that in the future individuals will depend to a great extent on the capacity to synthesise large amounts of information. ‘They will need to be able to gather together information from disparate sources and put it together in ways that work for themselves and can be communicated to other persons’(Gardner 2008, p. xiii). One of the first steps in ‘putting things together’ so they ‘work’ in the mind is ‘to group objects and events together on the basis of some similarity between them’ (Lee & das Gupta 1995, p. 116). When we do this and give them a collective name, we are conceptualising. Apart from helping to save our sanity by simplifying the vast amounts of data we encounter every day, concepts help us to understand and gain meaning from what we experience. Concepts are essential for synthesising information and they also help us to communicate with others. Put simply, concepts serve as building blocks for knowledge, understanding and communication. This chapter addresses the importance of teaching and learning about concepts and conceptual development in studies of society and environment. It proceeds as follows: first, it considers how individuals use concepts, and, second, it explores the characteristics of concepts; the third section presents a discussion of approaches that might be adopted by teachers intending to help their students build concepts in the classroom.

Ontology-based image annotation

With regard to the long-standing problem of the semantic gap between low-level image features and high-level human knowledge, the image retrieval community has recently shifted its emphasis from low-level features analysis to high-level image semantics extrac- tion. User studies reveal that users tend to seek information using high-level semantics. Therefore, image semantics extraction is of great importance to content-based image retrieval because it allows the users to freely express what images they want. Semantic content annotation is the basis for semantic content retrieval. The aim of image anno- tation is to automatically obtain keywords that can be used to represent the content of images. The major research challenges in image semantic annotation are: what is the basic unit of semantic representation? how can the semantic unit be linked to high-level image knowledge? how can the contextual information be stored and utilized for image annotation? In this thesis, the Semantic Web technology (i.e. ontology) is introduced to the image semantic annotation problem. Semantic Web, the next generation web, aims at mak- ing the content of whatever type of media not only understandable to humans but also to machines. Due to the large amounts of multimedia data prevalent on the Web, re- searchers and industries are beginning to pay more attention to the Multimedia Semantic Web. The Semantic Web technology provides a new opportunity for multimedia-based applications, but the research in this area is still in its infancy. Whether ontology can be used to improve image annotation and how to best use ontology in semantic repre- sentation and extraction is still a worth-while investigation. This thesis deals with the problem of image semantic annotation using ontology and machine learning techniques in four phases as below. 1) Salient object extraction. A salient object servers as the basic unit in image semantic extraction as it captures the common visual property of the objects. Image segmen- tation is often used as the �rst step for detecting salient objects, but most segmenta- tion algorithms often fail to generate meaningful regions due to over-segmentation and under-segmentation. We develop a new salient object detection algorithm by combining multiple homogeneity criteria in a region merging framework. 2) Ontology construction. Since real-world objects tend to exist in a context within their environment, contextual information has been increasingly used for improving object recognition. In the ontology construction phase, visual-contextual ontologies are built from a large set of fully segmented and annotated images. The ontologies are composed of several types of concepts (i.e. mid-level and high-level concepts), and domain contextual knowledge. The visual-contextual ontologies stand as a user-friendly interface between low-level features and high-level concepts. 3) Image objects annotation. In this phase, each object is labelled with a mid-level concept in ontologies. First, a set of candidate labels are obtained by training Support Vectors Machines with features extracted from salient objects. After that, contextual knowledge contained in ontologies is used to obtain the �nal labels by removing the ambiguity concepts. 4) Scene semantic annotation. The scene semantic extraction phase is to get the scene type by using both mid-level concepts and domain contextual knowledge in ontologies. Domain contextual knowledge is used to create scene con�guration that describes which objects co-exist with which scene type more frequently. The scene con�guration is represented in a probabilistic graph model, and probabilistic inference is employed to calculate the scene type given an annotated image. To evaluate the proposed methods, a series of experiments have been conducted in a large set of fully annotated outdoor scene images. These include a subset of the Corel database, a subset of the LabelMe dataset, the evaluation dataset of localized semantics in images, the spatial context evaluation dataset, and the segmented and annotated IAPR TC-12 benchmark.

Effectiveness of constructability concept in the provision of infrastructure assets

The concept of constructability is to use construction knowledge and experience during all phases of a project, particularly in the earliest phases of planning and design. It facilitates project objectives before delivery stage, and decreases unnecessary costs during construction phase. Despite the extensive use, constructability concept fails to address many issues related to Operation and Maintenance (O&M) of construction projects. Extending constructability concept, to include the O&M issues, could lead to the projects that are not fitted for construction purposes only, but also fitted for use. This study reviews the literature of constructability implementation, its benefits and shortcomings during the infrastructure life cycle, as well as the delivery success factors of infrastructure projects. This contributes to the propose need of a model to improve the effectiveness and efficiency of infrastructure project by extending the concept of constructability to include O&M. Development of such a model can facilitate post-occupancy stakeholders’ participation in a constructability program. It will help infrastructure owners eliminate project reworks, and improve O&M effectiveness and efficiency.

The variant concept of transportation disadvantaged : evidence from Aydin, Turkey and Yamaga, Japan

Transportation disadvantaged groups, in the previous studies, are defined as those who are low income earners, family dependent, limited access to private motor vehicles and public transport services, and also those who oblige to spend relatively more time and money on their trips. Additionally those disable, young and elderly are considered among the natural groups of transportation disadvantaged. Although in general terms this definition seems correct, it is not specific enough to become a common universal definition that could apply to all urban contexts. This paper investigates whether travel difficulty perceptions vary and so does the definition of transportation disadvantaged in socio-culturally different urban contexts. For this investigation the paper undertakes a series of statistical analysis in the case study of Yamaga, Japan, and compares the findings with a previous case study, where the same methodology, hypothesis, and assumptions were utilized in a culturally and demographically different settlement of Aydin, Turkey. After comparing the findings observed in Aydin with the statistical analysis results of Yamaga, this paper reveals that there can be no explicitly detailed universal definition of transportation disadvantaged. The paper concludes by stating characteristics of transportation disadvantage is not globally identical, and policies and solutions that work in a locality may not show the same results in another socio-cultural context.

The concept of 'Organismic Asymmetry' in sport science

The concept of organismic asymmetry refers to an inherent bias for seeking explanations of human performance and behaviour based on internal mechanisms and referents. A weakness in this tendency is a failure to consider the performer–environment relationship as the relevant scale of analysis. In this paper we elucidate the philosophical roots of the bias and discuss implications of organismic asymmetry for sport science and performance analysis, highlighting examples in psychology, sports medicine and biomechanics.

Extending constructability concept to include operation and maintenance issues

The concept of constructability uses integration art of individual functions through a valuable and timely construction inputs into planning and design development stages. It results in significant savings in cost and time needed to finalize infrastructure projects. However, available constructability principles, developed by CII Australia (1993), do not cover Operation and Maintenance (O&M) phases of projects, whilst major cost and time in multifaceted infrastructure projects are spent in post-occupancy stages. This paper discusses the need to extend the constructability concept by examining current O&M issues in the provision of multifaceted building projects. It highlights available O&M problems and shortcomings of building projects, as well as their causes and reasons in different categories. This initial categorization is an efficient start point for testing probable present O&M issues in various cases of complex infrastructure building projects. This preliminary categorization serve as a benchmark to develop an extended constructability model that considers the whole project life cycle phases rather than a specific phase. It anticipates that the development of an extended constructability model can reduce significant number of reworks, mistakes, extra costs and time wasted during delivery stages of multifaceted building projects.