Fast classification in incrementally growing spaces

[EN]The classification speed of state-of-the-art classifiers such as SVM is an important aspect to be considered for emerging applications and domains such as data mining and human-computer interaction. Usually, a test-time speed increase in SVMs is achieved by somehow reducing the number of support vectors, which allows a faster evaluation of the decision function. In this paper a novel approach is described for fast classification in a PCA+SVM scenario. In the proposed approach, classification of an unseen sample is performed incrementally in increasingly larger feature spaces. As soon as the classification confidence is above a threshold the process stops and the class label is retrieved...

Authentification biométrique par dynamique de frappe pour évaluation à distance utilisant SVM à une classe

La machine à vecteurs de support à une classe est un algorithme non-supervisé qui est capable d’apprendre une fonction de décision à partir de données d’une seule classe pour la détection d’anomalie. Avec les données d’entraînement d’une seule classe, elle peut identifier si une nouvelle donnée est similaire à l’ensemble d’entraînement. Dans ce mémoire, nous nous intéressons à la reconnaissance de forme de dynamique de frappe par la machine à vecteurs de support à une classe, pour l’authentification d’étudiants dans un système d’évaluation sommative à distance à l’Université Laval. Comme chaque étudiant à l’Université Laval possède un identifiant court, unique qu’il utilise pour tout accès sécurisé aux ressources informatiques, nous avons choisi cette chaîne de caractères comme support à la saisie de dynamique de frappe d’utilisateur pour construire notre propre base de données. Après avoir entraîné un modèle pour chaque étudiant avec ses données de dynamique de frappe, on veut pouvoir l’identifier et éventuellement détecter des imposteurs. Trois méthodes pour la classification ont été testées et discutées. Ainsi, nous avons pu constater les faiblesses de chaque méthode dans ce système. L’évaluation des taux de reconnaissance a permis de mettre en évidence leur dépendance au nombre de signatures ainsi qu’au nombre de caractères utilisés pour construire les signatures. Enfin, nous avons montré qu’il existe des corrélations entre le taux de reconnaissance et la dispersion dans les distributions des caractéristiques des signatures de dynamique de frappe.

An actor-critic algorithm with function approximation for discounted cost constrained Markov decision processes

We develop in this article the first actor-critic reinforcement learning algorithm with function approximation for a problem of control under multiple inequality constraints. We consider the infinite horizon discounted cost framework in which both the objective and the constraint functions are suitable expected policy-dependent discounted sums of certain sample path functions. We apply the Lagrange multiplier method to handle the inequality constraints. Our algorithm makes use of multi-timescale stochastic approximation and incorporates a temporal difference (TD) critic and an actor that makes a gradient search in the space of policy parameters using efficient simultaneous perturbation stochastic approximation (SPSA) gradient estimates. We prove the asymptotic almost sure convergence of our algorithm to a locally optimal policy. (C) 2010 Elsevier B.V. All rights reserved.

An Online Actor-Critic Algorithm with Function Approximation for Constrained Markov Decision Processes

We develop an online actor-critic reinforcement learning algorithm with function approximation for a problem of control under inequality constraints. We consider the long-run average cost Markov decision process (MDP) framework in which both the objective and the constraint functions are suitable policy-dependent long-run averages of certain sample path functions. The Lagrange multiplier method is used to handle the inequality constraints. We prove the asymptotic almost sure convergence of our algorithm to a locally optimal solution. We also provide the results of numerical experiments on a problem of routing in a multi-stage queueing network with constraints on long-run average queue lengths. We observe that our algorithm exhibits good performance on this setting and converges to a feasible point.

A novel Q-learning algorithm with function approximation for constrained Markov decision processes

We present a novel multi-timescale Q-learning algorithm for average cost control in a Markov decision process subject to multiple inequality constraints. We formulate a relaxed version of this problem through the Lagrange multiplier method. Our algorithm is different from Q-learning in that it updates two parameters - a Q-value parameter and a policy parameter. The Q-value parameter is updated on a slower time scale as compared to the policy parameter. Whereas Q-learning with function approximation can diverge in some cases, our algorithm is seen to be convergent as a result of the aforementioned timescale separation. We show the results of experiments on a problem of constrained routing in a multistage queueing network. Our algorithm is seen to exhibit good performance and the various inequality constraints are seen to be satisfied upon convergence of the algorithm.

Trademarks with a reputation:bridging the 'logical lapse'

The article focuses on the Trade Marks Regulations 2004, which came into force on May 5, 2004. Regulation 7 repeals s 5(3)(b) and amends s 10(3) of the Trade Marks Act 1994, implementing the decision of the European Court of Justice in Davidoff & Cie SA and Zino Davidoff SA v Gofkid Ltd., which was then confirmed in Adidas-Salomon AG and Adidas Benelux BV v Fitnessworld Trading Ltd. The orthodox definition of the primary and proper function of a trademark is that it is to identify the origin or ownership of the goods to which the mark is affixed. As society has changed so too have the functions of the trademark. Due to increased affluence and the growth of the consumer society, some trademarks may now have achieved the status of being a symbol of desire in their own right.

Optimal Decision Rules in Logical Recognition Models

The task of smooth and stable decision rules construction in logical recognition models is considered. Logical regularities of classes are defined as conjunctions of one-place predicates that determine the membership of features values in an intervals of the real axis. The conjunctions are true on a special no extending subsets of reference objects of some class and are optimal. The standard approach of linear decision rules construction for given sets of logical regularities consists in realization of voting schemes. The weighting coefficients of voting procedures are done as heuristic ones or are as solutions of complex optimization task. The modifications of linear decision rules are proposed that are based on the search of maximal estimations of standard objects for their classes and use approximations of logical regularities by smooth sigmoid functions.

Generalized Discernibility Function Based Attribute Reduction in Incomplete Decision Systems

A rough set approach for attribute reduction is an important research subject in data mining and machine learning. However, most attribute reduction methods are performed on a complete decision system table. In this paper, we propose methods for attribute reduction in static incomplete decision systems and dynamic incomplete decision systems with dynamically-increasing and decreasing conditional attributes. Our methods use generalized discernibility matrix and function in tolerance-based rough sets.

Combined quality function deployment and logical framework analysis to improve quality of emergency care in Malta

Purpose – The purpose of this paper is to develop an integrated patient-focused analytical framework to improve quality of care in accident and emergency (A&E) unit of a Maltese hospital. Design/methodology/approach – The study adopts a case study approach. First, a thorough literature review has been undertaken to study the various methods of healthcare quality management. Second, a healthcare quality management framework is developed using combined quality function deployment (QFD) and logical framework approach (LFA). Third, the proposed framework is applied to a Maltese hospital to demonstrate its effectiveness. The proposed framework has six steps, commencing with identifying patients’ requirements and concluding with implementing improvement projects. All the steps have been undertaken with the involvement of the concerned stakeholders in the A&E unit of the hospital. Findings – The major and related problems being faced by the hospital under study were overcrowding at A&E and shortage of beds, respectively. The combined framework ensures better A&E services and patient flow. QFD identifies and analyses the issues and challenges of A&E and LFA helps develop project plans for healthcare quality improvement. The important outcomes of implementing the proposed quality improvement programme are fewer hospital admissions, faster patient flow, expert triage and shorter waiting times at the A&E unit. Increased emergency consultant cover and faster first significant medical encounter were required to start addressing the problems effectively. Overall, the combined QFD and LFA method is effective to address quality of care in A&E unit. Practical/implications – The proposed framework can be easily integrated within any healthcare unit, as well as within entire healthcare systems, due to its flexible and user-friendly approach. It could be part of Six Sigma and other quality initiatives. Originality/value – Although QFD has been extensively deployed in healthcare setup to improve quality of care, very little has been researched on combining QFD and LFA in order to identify issues, prioritise them, derive improvement measures and implement improvement projects. Additionally, there is no research on QFD application in A&E. This paper bridges these gaps. Moreover, very little has been written on the Maltese health care system. Therefore, this study contributes demonstration of quality of emergency care in Malta.

International practices in investment decision making in road sector

This document provides the findings of an international review of investment decision-making practices in road asset management. Efforts were concentrated on identifying the strategic objectives of agencies in road asset management, establishing and understanding criteria different organisations adopted and ascertaining the exact methodologies used by different countries and international organisations. Road assets are powerful drivers of economic development and social equity. They also have significant impacts on the natural and man-made environment. The traditional definition of asset management is “A systematic process of maintaining, upgrading and operating physical assets cost effectively. It combines engineering principles with sound business practices and economic theory and it provides tools to facilitate a more organised, logical approach to decision-making” (US Dept. of Transportation, 1999). In recent years, the concept has been broadened to cover the complexity of decision making, based on a wider variety of policy considerations as well as social and environmental issues rather than is covered by Benefit-Cost analysis and pure technical considerations. Current international practices are summarised in table 2. It was evident that Engineering-economic analysis methods are well advanced to support decision-making. A range of tools available supports performance predicting of road assets and associated cost/benefit in technical context. The need for considering triple plus one bottom line of social, environmental and economic as well as political factors in decision-making is well understood by road agencies around the world. The techniques used to incorporate these however, are limited. Most countries adopt a scoring method, a goal achievement matrix or information collected from surveys. The greater uncertainty associated with these non-quantitative factors has generally not been taken into consideration. There is a gap between the capacities of the decision-making support systems and the requirements from decision-makers to make more rational and transparent decisions. The challenges faced in developing an integrated decision making framework are both procedural and conceptual. In operational terms, the framework should be easy to be understood and employed. In philosophical terms, the framework should be able to deal with challenging issues, such as uncertainty, time frame, network effects, model changes, while integrating cost and non-cost values into the evaluation. The choice of evaluation techniques depends on the feature of the problem at hand, on the aims of the analysis, and on the underlying information base At different management levels, the complexity in considering social, environmental, economic and political factor in decision-making is different. At higher the strategic planning level, more non-cost factors are involved. The complexity also varies based on the scope of the investment proposals. Road agencies traditionally place less emphasis on evaluation of maintenance works. In some cases, social equity, safety, environmental issues have been used in maintenance project selection. However, there is not a common base for the applications.

The role of decision-maker preferences in tenancy selection of CBD office accommodation – preliminary literature review

The literature and anecdotal evidence suggests that that there is more to tenancy selection (firm location) than the profit maximisation drive that traditional neo-classical economic location theory suggests. In the first instance these models assume property markets are rational and perfectly competitive; the CBD office market is clearly neither rational nor perfectly competitive. This fact alone relegates such models to the margins of usefulness for an industry that seeks to satisfy tenant demand in order to optimise returns on capital invested. Acknowledgment of property market imperfections are universally accepted to the extent that all contemporary texts discuss the lack of a coherent centralised market place and incomplete and poorly disseminated information processes as fundamental inadequacies which characterise the property market inefficiencies. Less well researched are the facets of the market which allow the observer to determine market activity to be significantly irrational. One such facet is that of ‘decision maker preferences’. The decision to locate a business operation at one location as opposed to another seems ostensibly a routine choice based on short, medium and long term business objectives. These objectives are derived from a process of strategic planning by one or more individuals whose goal is held to be to optimise outcomes which benefit the business (and presumably those employed within it). However the decision making processes appear bounded by how firms function, the institutional context in which they operate, as well as by opportunistic behaviour by individual decision makers who allow personal preferences to infiltrate and ‘corrupt’ the process. In this way, history, culture, geography, as well as institutions all become significant to the extent that these influence and shape individual behaviour which in turn determine the morphology of individual preferences, as well as providing a conduit for them to take effect. This paper exams historical and current literature on the impact of individual behaviour in the decision making process within organisations as a precursor to an investigation of the tenancy decision making process within the CBD office market. Literature on the topic falls within a number of research disciplines, philosophy, psychology and economics to name a few.

Remaining useful life prediction using elliptical basis function network and Markov Chain

This paper presents a novel method for remaining useful life prediction using the Elliptical Basis Function (EBF) network and a Markov chain. The EBF structure is trained by a modified Expectation-Maximization (EM) algorithm in order to take into account the missing covariate set. No explicit extrapolation is needed for internal covariates while a Markov chain is constructed to represent the evolution of external covariates in the study. The estimated external and the unknown internal covariates constitute an incomplete covariate set which are then used and analyzed by the EBF network to provide survival information of the asset. It is shown in the case study that the method slightly underestimates the remaining useful life of an asset which is a desirable result for early maintenance decision and resource planning.

Physical function and health-related quality of life of older adults undergoing hospital rehabilitation : how strong is the association?

The purpose of this research is to report preliminary empirical evidence regarding the association between common physical performance measures and health-related quality of life (HRQoL) of hospitalized older adults recovering from illness and injury. Frequently, these patients do not return to premorbid levels of independence and physical ability. Rehabilitation for this population often focuses on improving physical functioning and mobility with the intention of maximizing their HRQoL for discharge and thereafter. For this reason, longitudinal use of physical performance measures as an indicator of improvement in physical functioning (and thus HRQoL) is common. Although this is a logical approach, there have been mixed results from previous investigations into the association between common measures of physical function and HRQoL amongst other adult patient populations.1,2 There has been no previous investigation reporting the association between HRQoL and a variety of common physical performance measures in hospitalized older adults.

Bounded parameter Markov Decision Processes with average reward criterion

Bounded parameter Markov Decision Processes (BMDPs) address the issue of dealing with uncertainty in the parameters of a Markov Decision Process (MDP). Unlike the case of an MDP, the notion of an optimal policy for a BMDP is not entirely straightforward. We consider two notions of optimality based on optimistic and pessimistic criteria. These have been analyzed for discounted BMDPs. Here we provide results for average reward BMDPs. We establish a fundamental relationship between the discounted and the average reward problems, prove the existence of Blackwell optimal policies and, for both notions of optimality, derive algorithms that converge to the optimal value function.