A Bayesian formulation of search, control and the exploration/exploitation trade-off

A new approach to optimisation is introduced based on a precise probabilistic statement of what is ideally required of an optimisation method. It is convenient to express the formalism in terms of the control of a stationary environment. This leads to an objective function for the controller which unifies the objectives of exploration and exploitation, thereby providing a quantitative principle for managing this trade-off. This is demonstrated using a variant of the multi-armed bandit problem. This approach opens new possibilities for optimisation algorithms, particularly by using neural network or other adaptive methods for the adaptive controller. It also opens possibilities for deepening understanding of existing methods. The realisation of these possibilities requires research into practical approximations of the exact formalism.

Social control and knowledge in democratic societies

In this article we introduce the notions of knowledge policy and the politics of knowledge. These have to be distinguished from the older, well-known terms of research policy, or science and technology policy. While the latter aim to foster the development of innovations in knowledge and its applications, the former is aware of side effects of new knowledge and tries to address them. While research policy takes the aims of innovations as largely unproblematic (insofar as they help improving national competitiveness), knowledge policy tries to govern (regulate, control, restrict, or even forbid) the production of knowledge.

"Well did you feel jealous?" Control and ideology in police interviews with rape complainants

The methods used by the UK Police to investigate complaints of rape have unsurprisingly come under much scrutiny in recent times, with a 2007 joint report on behalf of HM Crown Prosecution Service Inspectorate and HM Inspectorate of Constabulary concluding that there were many areas where improvements should be made. The research reported here forms part of a larger project which draws on various discourse analytical tools to identify the processes at work during police interviews with women reporting rape. Drawing on a corpus of video recorded police interviews with women reporting rape, this study applies a two pronged analysis to reveal the presence of these ideologies. Firstly, an analysis of the discourse markers ‘well’ and ‘so’ demonstrates the control exerted on the interaction by interviewing officers, as they attach importance to certain facts while omitting much of the information provided by the victim. Secondly, the interpretative repertoires relied upon by officers to ‘make sense’ of victim’s accounts are subject to scrutiny. As well as providing micro-level analyses which demonstrate processes of interactional control at the local level, the findings of these analyses can be shown to relate to a wider context – specifically prevailing ideologies about sexual violence in society as a whole.

The role of leader-member exchanges in mediating the relationship between locus of control and work reactions

The relationship between locus of control, the quality of exchanges between subordinates and leaders (LMX), and a variety of work-related reactions (intrinsic/extrinsic job satisfaction, work-related well-being, and organizational commitment) are examined. It was predicted that people with an internal locus of control develop better quality relations with their manager and this, in turn, results in more favourable work-related reactions. Results from two different samples (N=404, and N=51) supported this prediction, and also showed that LMX either fully, or partially, mediated the relationship between locus of control and all the work-related reactions.

Direct real-time cavity mode assessment and optimization technique for fiber lasers

A simple technique for direct real-time assessment of a fiber laser cavity-mode condition during operation is demonstrated. Mode stabilization and optimization with this cavity-mode monitoring and conditioning feedback scheme shows significant improvements to the output performance.

Inference and optimization of real edges on sparse graphs:A statistical physics perspective

Inference and optimization of real-value edge variables in sparse graphs are studied using the Bethe approximation and replica method of statistical physics. Equilibrium states of general energy functions involving a large set of real edge variables that interact at the network nodes are obtained in various cases. When applied to the representative problem of network resource allocation, efficient distributed algorithms are also devised. Scaling properties with respect to the network connectivity and the resource availability are found, and links to probabilistic Bayesian approximation methods are established. Different cost measures are considered and algorithmic solutions in the various cases are devised and examined numerically. Simulation results are in full agreement with the theory. © 2007 The American Physical Society.