Closing the gap between bandit and full-information online optimization : high-probability regret bound

We demonstrate a modification of the algorithm of Dani et al for the online linear optimization problem in the bandit setting, which allows us to achieve an O( \sqrt{T ln T} ) regret bound in high probability against an adaptive adversary, as opposed to the in expectation result against an oblivious adversary of Dani et al. We obtain the same dependence on the dimension as that exhibited by Dani et al. The results of this paper rest firmly on those of Dani et al and the remarkable technique of Auer et al for obtaining high-probability bounds via optimistic estimates. This paper answers an open question: it eliminates the gap between the high-probability bounds obtained in the full-information vs bandit settings.

High-probability regret bounds for bandit online linear optimization

We present a modification of the algorithm of Dani et al. [8] for the online linear optimization problem in the bandit setting, which with high probability has regret at most O ∗ ( √ T) against an adaptive adversary. This improves on the previous algorithm [8] whose regret is bounded in expectation against an oblivious adversary. We obtain the same dependence on the dimension (n 3/2) as that exhibited by Dani et al. The results of this paper rest firmly on those of [8] and the remarkable technique of Auer et al. [2] for obtaining high probability bounds via optimistic estimates. This paper answers an open question: it eliminates the gap between the high-probability bounds obtained in the full-information vs bandit settings.

Stability and change

This special feature section of Journal of Management & Organization (Volume 17/1 - March 2011) sets out to widen understanding of the processes of stability and change in today's organizations, with a particular emphasis on the contribution of institutional approaches to organizational studies. Institutional perspectives on organization theory assume that rational, economic calculations, such as the maximization of profits or the optimization of resource allocation, are not sufficient to understand the behavior of organizations and their strategic choices. Institutionalists acknowledge the great uncertainty associated with the conduct of organizations and suggest that taken-for-granted values, beliefs and meanings within and outside organizations also play an important role in the determination of legitimate action.

Assessment of requirements for establishment of a framework to enhance implementation of quality practices in building projects

Quality, in construction projects should be regarded as the fulfillment of expectation of those contributors involved in such projects. Although a significant amount of quality practices have been introduced within the industry, attainment of reasonable levels of quality in construction projects continues to be an ongoing problem. To date, some research into the introduction and improvement of quality practices and stakeholder management has been undertaken, but so far no major studies have been completed that comprehensively examine how greater consideration of stakeholders’ perspectives of quality can be used to contribute to final project quality outcomes. This paper aims to examine the requirements for development of a framework leading to more effective involvement of stakeholders in quality planning and practices thus ultimately contributing to higher quality outcomes for construction projects. Through an extensive literature review it highlights various perceptions of quality, categorizes quality issues with particular focus on benefits and shortcomings and also examines the viewpoints of major stakeholders on project quality. It proposes a set of criteria to be used as a basis for a quality practice improvement framework, which will provide project managers and owners with the required information and strategic direction to achieve their own and their stakeholders’ targets for implementation of quality practices leading to the achievement of improved quality outcomes on future projects.

Aggregate distance based clustering using Fibonacci Series -FIBCLUS

This paper proposes an innovative instance similarity based evaluation metric that reduces the search map for clustering to be performed. An aggregate global score is calculated for each instance using the novel idea of Fibonacci series. The use of Fibonacci numbers is able to separate the instances effectively and, in hence, the intra-cluster similarity is increased and the inter-cluster similarity is decreased during clustering. The proposed FIBCLUS algorithm is able to handle datasets with numerical, categorical and a mix of both types of attributes. Results obtained with FIBCLUS are compared with the results of existing algorithms such as k-means, x-means expected maximization and hierarchical algorithms that are widely used to cluster numeric, categorical and mix data types. Empirical analysis shows that FIBCLUS is able to produce better clustering solutions in terms of entropy, purity and F-score in comparison to the above described existing algorithms.

Citizen Coombs Wins!

Citizen Coombs Wins! appears to be a standard arcade game placed within the gallery. Mortal Kombat is displayed on the screen inviting the viewer to press play. The ‘player’ selects their character and awaits the commencement of the game; at first move however, the player dies – sound and text informs them that ‘Citizen Coombs Wins!’. By altering the expected play of the game, this work exploring notions of play, control, the institution and expectation. This work seeks to invite, engage and repel the viewer in order to question, critique and play with the role of the artist and the viewer within the context of the institution. The work was included in the international group show 'Ceci n'est pas une Casino!', curated by Kevin Muhlen and Jo Kox for the Casino Luxembourg and later toured to Villa Merkel, Esslingen, Germany.

Practical decision support system evaluation for consultant selection

The widespread development of Decision Support System (DSS) in construction indicate that the evaluation of software become more important than before. However, it is identified that most research in construction discipline did not attempt to assess its usability. Therefore, little is known about the approach on how to properly evaluate a DSS for specific problem. In this paper, we present a practical framework that can be guidance for DSS evaluation. It focuses on how to evaluate software that is dedicatedly designed for consultant selection problem. The framework features two main components i.e. Sub-system Validation and Face Validation. Two case studies of consultant selection at Malaysian Department of Irrigation and Drainage were integrated in this framework. Some inter-disciplinary area such as Software Engineering, Human Computer Interaction (HCI) and Construction Project Management underpinned the discussion of the paper. It is anticipated that this work can foster better DSS development and quality decision making that accurately meet the client’s expectation and needs

Educational stress scale for adolescents : development, validity, and reliability with Chinese students

This article describes the development and initial validation of a new instrument to measure academic stress—the Educational Stress Scale for Adolescents (ESSA). A series of cross-sectional questionnaire surveys were conducted with more than 2,000 Chinese adolescents to examine the psychometric properties. The final 16-item ESSA contains five latent variables: Pressure from study, Workload, Worry about grades, Self-expectation, and Despondency, which together explain 64% of the total item variance. Scale scores showed adequate internal consistency, 2-week test–retest reliability, and satisfactory concurrent validity. A confirmatory factor analysis suggested the proposed factor model fits well in a different sample. For researchers who have a particular interest in academic stress among adolescents, the ESSA promises to be a useful tool.

Boardroom identities: can independence by 'cued' in directors?

Director independence is a cornerstone of fiduciary responsibility and good corporate governance. However, most directors are recruited because of the roles and networks they hold, meaning that there is an expectation that identities held by a director outside the boardroom will be used to benefit the company. While this often works well, it is acknowledged that many directors, either consciously or subconsciously, will at times allow themselves to be influenced by their other roles to the detriment of the governance process. In this paper we argue that identity theory can be used to explore the impact of ‘identity’ on corporate governance and that practical tools can be developed to actively assist directors to maintain ‘independence’ in the boardroom.

Verifiability in Markets for Credence Goods

Theory predicts that efficiency prevails on credence goods markets if customers are able to verify which quality they receive from an expert seller. In a series of experiments with endogenous prices we observe that variability fails to result in efficient provision behavior and leads to very similar results as a setting without variability. Some sellers always provide appropriate treatment even if own money maximization calls for over- or undertreatment. Overall our endogenous price-results suggests that both inequality aversion and a taste for efficiency play an important role for experts provision behavior. We contrast the implications of those two motivations theoretically and discriminate between them empirically using a �xed-price design. We then classify experimental experts according to their provision behavior.

A superdirective 3-element array for adaptive beamforming

A small array composed of three monopole elements with very small element spacing on the order of λ/6 to λ/20 is considered for application in adaptive beamforming. The properties of this 3-port array are governed by strong mutual coupling. It is shown that for signal-to-noise maximization, it is not sufficient to adjust the weights to compensate for the effects of mutual coupling. The necessity for a RF-decoupling network (RF-DN) and its simple realization are shown. The array with closely spaced elements together with the RF-DN represents a superdirective antenna with a directivity of more than 10 dBi. It is shown that the required fractional frequency bandwidth and the available unloaded Q of the antenna and RF-DN structure determine the lower limit for the element spacing.

Investigating pre-service teachers’ perceptions of their preparedness to teach

The experiences of transition to the teaching profession have a significant impact on a teachers’ potential length of career, feelings of professional efficacy and the quality of performance in the classroom (Gore & Thomas, 2003). While the transition to practice is characterized by much expectation and excitement, it also a time of stress and uncertainty for many beginning teachers. As such, it is important to investigate this period of transition for beginning teachers. This paper explores graduate teachers perceptions of their personal ‘preparedness to teach’. The group is graduating from one Australian university, and the data is captured at the end of their teacher preparation programs,before they take up positions in schools. These graduating pre-service teachers are from one year graduate entry programs that include individual programs of early years, middle years and senior years. The key findings indicate that this group of graduating pre-service teachers are already engaged in some level of reflective practice and are actively seeking further professional learning to improve the practical aspects of their classroom teaching.

Assessing financial reporting comparability across institutional settings : the case of pension accounting

The drive for comparability of financial information is to enable users to distinguish similarities and differences in economic activities for an entity over time and between entities so that their resource allocation decisions are facilitated. With the increased globalisation of economic activities, the enhanced international comparability of financial statements is often used as an argument to advance the convergence of local accounting standards to international financial reporting standards (IFRS). Differences in the underlying economic substance of transactions between jurisdictions plus accounting standards allowing alternative treatments may render this expectation of increased comparability unrealistic. Motivated by observations that, as a construct, comparability is under-researched and not well understood, we develop a comparability framework that distinguishes between four types of comparability. In applying this comparability framework to pension accounting in the Australian and USA contexts, we highlight a dilemma: while regulators seek to increase the likelihood that similar events are accounted for similarly, an unintended consequence may be that preparers are forced to apply similar accounting treatment to events that are, in substance, different.

Quality metrics for assessing security-critical computer programs

Existing secure software development principles tend to focus on coding vulnerabilities, such as buffer or integer overflows, that apply to individual program statements, or issues associated with the run-time environment, such as component isolation. Here we instead consider software security from the perspective of potential information flow through a program’s object-oriented module structure. In particular, we define a set of quantifiable "security metrics" which allow programmers to quickly and easily assess the overall security of a given source code program or object-oriented design. Although measuring quality attributes of object-oriented programs for properties such as maintainability and performance has been well-covered in the literature, metrics which measure the quality of information security have received little attention. Moreover, existing securityrelevant metrics assess a system either at a very high level, i.e., the whole system, or at a fine level of granularity, i.e., with respect to individual statements. These approaches make it hard and expensive to recognise a secure system from an early stage of development. Instead, our security metrics are based on well-established compositional properties of object-oriented programs (i.e., data encapsulation, cohesion, coupling, composition, extensibility, inheritance and design size), combined with data flow analysis principles that trace potential information flow between high- and low-security system variables. We first define a set of metrics to assess the security quality of a given object-oriented system based on its design artifacts, allowing defects to be detected at an early stage of development. We then extend these metrics to produce a second set applicable to object-oriented program source code. The resulting metrics make it easy to compare the relative security of functionallyequivalent system designs or source code programs so that, for instance, the security of two different revisions of the same system can be compared directly. This capability is further used to study the impact of specific refactoring rules on system security more generally, at both the design and code levels. By measuring the relative security of various programs refactored using different rules, we thus provide guidelines for the safe application of refactoring steps to security-critical programs. Finally, to make it easy and efficient to measure a system design or program’s security, we have also developed a stand-alone software tool which automatically analyses and measures the security of UML designs and Java program code. The tool’s capabilities are demonstrated by applying it to a number of security-critical system designs and Java programs. Notably, the validity of the metrics is demonstrated empirically through measurements that confirm our expectation that program security typically improves as bugs are fixed, but worsens as new functionality is added.

'Nothing exists for one cause' : putting biology back into evolution

The opening phrase of the title is from Charles Darwin’s notebooks (Schweber 1977). It is a double reminder, firstly that mainstream evolutionary theory is not just about describing nature but is particularly looking for mechanisms or ‘causes’, and secondly, that there will usually be several causes affecting any particular outcome. The second part of the title is our concern at the almost universal rejection of the idea that biological mechanisms are sufficient for macroevolutionary changes, thus rejecting a cornerstone of Darwinian evolutionary theory. Our primary aim here is to consider ways of making it easier to develop and to test hypotheses about evolution. Formalizing hypotheses can help generate tests. In an absolute sense, some of the discussion by scientists about evolution is little better than the lack of reasoning used by those advocating intelligent design. Our discussion here is in a Popperian framework where science is defined by that area of study where it is possible, in principle, to find evidence against hypotheses – they are in principle falsifiable. However, with time, the boundaries of science keep expanding. In the past, some aspects of evolution were outside the current boundaries of falsifiable science, but increasingly new techniques and ideas are expanding the boundaries of science and it is appropriate to re-examine some topics. It often appears that over the last few decades there has been an increasingly strong assumption to look first (and only) for a physical cause. This decision is virtually never formally discussed, just an assumption is made that some physical factor ‘drives’ evolution. It is necessary to examine our assumptions much more carefully. What is meant by physical factors ‘driving’ evolution, or what is an ‘explosive radiation’. Our discussion focuses on two of the six mass extinctions, the fifth being events in the Late Cretaceous, and the sixth starting at least 50,000 years ago (and is ongoing). Cretaceous/Tertiary boundary; the rise of birds and mammals. We have had a long-term interest (Cooper and Penny 1997) in designing tests to help evaluate whether the processes of microevolution are sufficient to explain macroevolution. The real challenge is to formulate hypotheses in a testable way. For example the numbers of lineages of birds and mammals that survive from the Cretaceous to the present is one test. Our first estimate was 22 for birds, and current work is tending to increase this value. This still does not consider lineages that survived into the Tertiary, and then went extinct later. Our initial suggestion was probably too narrow in that it lumped four models from Penny and Phillips (2004) into one model. This reduction is too simplistic in that we need to know about survival and ecological and morphological divergences during the Late Cretaceous, and whether Crown groups of avian or mammalian orders may have existed back into the Cretaceous. More recently (Penny and Phillips 2004) we have formalized hypotheses about dinosaurs and pterosaurs, with the prediction that interactions between mammals (and groundfeeding birds) and dinosaurs would be most likely to affect the smallest dinosaurs, and similarly interactions between birds and pterosaurs would particularly affect the smaller pterosaurs. There is now evidence for both classes of interactions, with the smallest dinosaurs and pterosaurs declining first, as predicted. Thus, testable models are now possible. Mass extinction number six: human impacts. On a broad scale, there is a good correlation between time of human arrival, and increased extinctions (Hurles et al. 2003; Martin 2005; Figure 1). However, it is necessary to distinguish different time scales (Penny 2005) and on a finer scale there are still large numbers of possibilities. In Hurles et al. (2003) we mentioned habitat modification (including the use of Geogenes III July 2006 31 fire), introduced plants and animals (including kiore) in addition to direct predation (the ‘overkill’ hypothesis). We need also to consider prey switching that occurs in early human societies, as evidenced by the results of Wragg (1995) on the middens of different ages on Henderson Island in the Pitcairn group. In addition, the presence of human-wary or humanadapted animals will affect the distribution in the subfossil record. A better understanding of human impacts world-wide, in conjunction with pre-scientific knowledge will make it easier to discuss the issues by removing ‘blame’. While continued spontaneous generation was accepted universally, there was the expectation that animals continued to reappear. New Zealand is one of the very best locations in the world to study many of these issues. Apart from the marine fossil record, some human impact events are extremely recent and the remains less disrupted by time.