Feature selection for clustering categorical data with an embedded modelling approach

Research on the problem of feature selection for clustering continues to develop. This is a challenging task, mainly due to the absence of class labels to guide the search for relevant features. Categorical feature selection for clustering has rarely been addressed in the literature, with most of the proposed approaches having focused on numerical data. In this work, we propose an approach to simultaneously cluster categorical data and select a subset of relevant features. Our approach is based on a modification of a finite mixture model (of multinomial distributions), where a set of latent variables indicate the relevance of each feature. To estimate the model parameters, we implement a variant of the expectation-maximization algorithm that simultaneously selects the subset of relevant features, using a minimum message length criterion. The proposed approach compares favourably with two baseline methods: a filter based on an entropy measure and a wrapper based on mutual information. The results obtained on synthetic data illustrate the ability of the proposed expectation-maximization method to recover ground truth. An application to real data, referred to official statistics, shows its usefulness.

Categorical data clustering using a minimum message length criterion

Research on cluster analysis for categorical data continues to develop, new clustering algorithms being proposed. However, in this context, the determination of the number of clusters is rarely addressed. We propose a new approach in which clustering and the estimation of the number of clusters is done simultaneously for categorical data. We assume that the data originate from a finite mixture of multinomial distributions and use a minimum message length criterion (MML) to select the number of clusters (Wallace and Bolton, 1986). For this purpose, we implement an EM-type algorithm (Silvestre et al., 2008) based on the (Figueiredo and Jain, 2002) approach. The novelty of the approach rests on the integration of the model estimation and selection of the number of clusters in a single algorithm, rather than selecting this number based on a set of pre-estimated candidate models. The performance of our approach is compared with the use of Bayesian Information Criterion (BIC) (Schwarz, 1978) and Integrated Completed Likelihood (ICL) (Biernacki et al., 2000) using synthetic data. The obtained results illustrate the capacity of the proposed algorithm to attain the true number of cluster while outperforming BIC and ICL since it is faster, which is especially relevant when dealing with large data sets.

Determining the number of clusters in categorical data

Cluster analysis for categorical data has been an active area of research. A well-known problem in this area is the determination of the number of clusters, which is unknown and must be inferred from the data. In order to estimate the number of clusters, one often resorts to information criteria, such as BIC (Bayesian information criterion), MML (minimum message length, proposed by Wallace and Boulton, 1968), and ICL (integrated classification likelihood). In this work, we adopt the approach developed by Figueiredo and Jain (2002) for clustering continuous data. They use an MML criterion to select the number of clusters and a variant of the EM algorithm to estimate the model parameters. This EM variant seamlessly integrates model estimation and selection in a single algorithm. For clustering categorical data, we assume a finite mixture of multinomial distributions and implement a new EM algorithm, following a previous version (Silvestre et al., 2008). Results obtained with synthetic datasets are encouraging. The main advantage of the proposed approach, when compared to the above referred criteria, is the speed of execution, which is especially relevant when dealing with large data sets.

New insights on integer-programming models for the kidney exchange problem

In recent years several countries have set up policies that allow exchange of kidneys between two or more incompatible patient–donor pairs. These policies lead to what is commonly known as kidney exchange programs. The underlying optimization problems can be formulated as integer programming models. Previously proposed models for kidney exchange programs have exponential numbers of constraints or variables, which makes them fairly difficult to solve when the problem size is large. In this work we propose two compact formulations for the problem, explain how these formulations can be adapted to address some problem variants, and provide results on the dominance of some models over others. Finally we present a systematic comparison between our models and two previously proposed ones via thorough computational analysis. Results show that compact formulations have advantages over non-compact ones when the problem size is large.

Legal and political regulation of higher education in Portugal: an evolutionary perspective on the current policy issues and the problem of order

The higher education system in Europe is currently under stress and the debates over its reform and future are gaining momentum. Now that, for most countries, we are in a time for change, in the overall society and the whole education system, the legal and political dimensions have gained prominence, which has not been followed by a more integrative approach of the problem of order, its reform and the issue of regulation, beyond the typical static and classical cost-benefit analyses. The two classical approaches for studying (and for designing the policy measures of) the problem of the reform of the higher education system - the cost-benefit analysis and the legal scholarship description - have to be integrated. This is the argument of our paper that the very integration of economic and legal approaches, what Warren Samuels called the legal-economic nexus, is meaningful and necessary, especially if we want to address the problem of order (as formulated by Joseph Spengler) and the overall regulation of the system. On the one hand, and without neglecting the interest and insights gained from the cost-benefit analysis, or other approaches of value for money assessment, we will focus our study on the legal, social and political aspects of the regulation of the higher education system and its reform in Portugal. On the other hand, the economic and financial problems have to be taken into account, but in a more inclusive way with regard to the indirect and other socio-economic costs not contemplated in traditional or standard assessments of policies for the tertiary education sector. In the first section of the paper, we will discuss the theoretical and conceptual underpinning of our analysis, focusing on the evolutionary approach, the role of critical institutions, the legal-economic nexus and the problem of order. All these elements are related to the institutional tradition, from Veblen and Commons to Spengler and Samuels. The second section states the problem of regulation in the higher education system and the issue of policy formulation for tackling the problem. The current situation is clearly one of crisis with the expansion of the cohorts of young students coming to an end and the recurrent scandals in private institutions. In the last decade, after a protracted period of extension or expansion of the system, i. e., the continuous growth of students, universities and other institutions are competing harder to gain students and have seen their financial situation at risk. It seems that we are entering a period of radical uncertainty, higher competition and a new configuration that is slowly building up is the growth in intensity, which means upgrading the quality of the higher learning and getting more involvement in vocational training and life-long learning. With this change, and along with other deep ones in the Portuguese society and economy, the current regulation has shown signs of maladjustment. The third section consists of our conclusions on the current issue of regulation and policy challenge. First, we underline the importance of an evolutionary approach to a process of change that is essentially dynamic. A special attention will be given to the issues related to an evolutionary construe of policy analysis and formulation. Second, the integration of law and economics, through the notion of legal economic nexus, allows us to better define the issues of regulation and the concrete problems that the universities are facing. One aspect is the instability of the political measures regarding the public administration and on which the higher education system depends financially, legally and institutionally, to say the least. A corollary is the lack of clear strategy in the policy reforms. Third, our research criticizes several studies, such as the one made by the OECD in late 2006 for the Ministry of Science, Technology and Higher Education, for being too static and neglecting fundamental aspects of regulation such as the logic of actors, groups and organizations who are major players in the system. Finally, simply changing the legal rules will not necessary per se change the behaviors that the authorities want to change. By this, we mean that it is not only remiss of the policy maker to ignore some of the critical issues of regulation, namely the continuous non-respect by academic management and administrative bodies of universities of the legal rules that were once promulgated. Changing the rules does not change the problem, especially without the necessary debates form the different relevant quarters that make up the higher education system. The issues of social interaction remain as intact. Our treatment of the matter will be organized in the following way. In the first section, the theoretical principles are developed in order to be able to study more adequately the higher education transformation with a modest evolutionary theory and a legal and economic nexus of the interactions of the system and the policy challenges. After describing, in the second section, the recent evolution and current working of the higher education in Portugal, we will analyze the legal framework and the current regulatory practices and problems in light of the theoretical framework adopted. We will end with some conclusions on the current problems of regulation and the policy measures that are discusses in recent years.

Reconhecimento de língua gestual

Trabalho Final de Mestrado para obtenção do grau de Mestre em Engenharia Informática e Computadores

Constraint solving over multi-valued logics - application to digital circuits

Due to usage conditions, hazardous environments or intentional causes, physical and virtual systems are subject to faults in their components, which may affect their overall behaviour. In a ‘black-box’ agent modelled by a set of propositional logic rules, in which just a subset of components is externally visible, such faults may only be recognised by examining some output function of the agent. A (fault-free) model of the agent’s system provides the expected output given some input. If the real output differs from that predicted output, then the system is faulty. However, some faults may only become apparent in the system output when appropriate inputs are given. A number of problems regarding both testing and diagnosis thus arise, such as testing a fault, testing the whole system, finding possible faults and differentiating them to locate the correct one. The corresponding optimisation problems of finding solutions that require minimum resources are also very relevant in industry, as is minimal diagnosis. In this dissertation we use a well established set of benchmark circuits to address such diagnostic related problems and propose and develop models with different logics that we formalise and generalise as much as possible. We also prove that all techniques generalise to agents and to multiple faults. The developed multi-valued logics extend the usual Boolean logic (suitable for faultfree models) by encoding values with some dependency (usually on faults). Such logics thus allow modelling an arbitrary number of diagnostic theories. Each problem is subsequently solved with CLP solvers that we implement and discuss, together with a new efficient search technique that we present. We compare our results with other approaches such as SAT (that require substantial duplication of circuits), showing the effectiveness of constraints over multi-valued logics, and also the adequacy of a general set constraint solver (with special inferences over set functions such as cardinality) on other problems. In addition, for an optimisation problem, we integrate local search with a constructive approach (branch-and-bound) using a variety of logics to improve an existing efficient tool based on SAT and ILP.

Quantifying the channel quality for interference-aware wireless sensor networks

Reliability of communications is key to expand application domains for sensor networks. SinceWireless Sensor Networks (WSN) operate in the license-free Industrial Scientific and Medical (ISM) bands and hence share the spectrum with other wireless technologies, addressing interference is an important challenge. In order to minimize its effect, nodes can dynamically adapt radio resources provided information about current spectrum usage is available. We present a new channel quality metric, based on availability of the channel over time, which meaningfully quantifies spectrum usage. We discuss the optimum scanning time for capturing the channel condition while maintaining energy-efficiency. Using data collected from a number of Wi-Fi networks operating in a library building, we show that our metric has strong correlation with the Packet Reception Rate (PRR). This suggests that quantifying interference in the channel can help in adapting resources for better reliability. We present a discussion of the usage of our metric for various resource allocation and adaptation strategies.

A fractional approach to the Fermi-Pasta-Ulam problem

This paper studies the Fermi-Pasta-Ulam problem having in mind the generalization provided by Fractional Calculus (FC). The study starts by addressing the classical formulation, based on the standard integer order differential calculus and evaluates the time and frequency responses. A first generalization to be investigated consists in the direct replacement of the springs by fractional elements of the dissipative type. It is observed that the responses settle rapidly and no relevant phenomena occur. A second approach consists of replacing the springs by a blend of energy extracting and energy inserting elements of symmetrical fractional order with amplitude modulated by quadratic terms. The numerical results reveal a response close to chaotic behaviour.

Notional processors: an approach for multiprocessor scheduling

Consider the problem of designing an algorithm with a high utilisation bound for scheduling sporadic tasks with implicit deadlines on identical processors. A task is characterised by its minimum interarrival time and its execution time. Task preemption and migration is permitted. Still, low preemption and migration counts are desirable. We formulate an algorithm with a utilisation bound no less than 66.¯6%, characterised by worst-case preemption counts comparing favorably against the state-of-the-art.

Two protocols without periodicity for the global and preemptive scheduling problem of multi-mode real-time systems upon multiprocessor Platforms

We consider the problem of scheduling a multi-mode real-time system upon identical multiprocessor platforms. Since it is a multi-mode system, the system can change from one mode to another such that the current task set is replaced with a new task set. Ensuring that deadlines are met requires not only that a schedulability test is performed on tasks in each mode but also that (i) a protocol for transitioning from one mode to another is specified and (ii) a schedulability test for each transition is performed. We propose two protocols which ensure that all the expected requirements are met during every transition between every pair of operating modes of the system. Moreover, we prove the correctness of our proposed algorithms by extending the theory about the makespan determination problem.

Real-time communications over wired/wireless PROFIBUS networks supporting inter-cell mobility

PROFIBUS is an international standard (IEC 61158, EN 50170) for factory-floor communications, with several thousands of installations worldwide. Taking into account the increasing need for mobile devices in industrial environments, one obvious solution is to extend traditional wired PROFIBUS networks with wireless capabilities. In this paper, we outline the major aspects of a hybrid wired/wireless PROFIBUS-based architecture, where most of the design options were made in order to guarantee the real-time behaviour of the overall network. We also introduce the timing unpredictability problems resulting from the co-existence of heterogeneous physical media in the same network. However, the major focus of this paper is on how to guarantee real-time communications in such a hybrid network, where nodes (and whole segments) can move between different radio cells (inter-cell mobility). Assuming a simple mobility management mechanism based on mobile nodes performing periodic radio channel assessment and switching, we propose a methodology to compute values for specific parameters that enable an optimal (minimum) and bounded duration of the handoff procedure.

WiDom: a dominance protocol for wireless medium access

In this paper, we address the problem of sharing a wireless channel among a set of sporadic message streams where a message stream issues transmission requests with real-time deadlines. We propose a collision-free wireless medium access control (MAC) protocol which implements static-priority scheduling, supports a large number of priority levels and is fully distributed. It is an adaptation to a wireless channel of the dominance protocol used in the CAN bus. But, unlike that protocol, our protocol does not require a node having the ability to receive an incoming bit from the channel while transmitting to the channel. The evaluation of the protocol with real embedded computing platforms is presented to show that the proposed protocol is in fact collision-free and prioritized. We measure the response times of our implementation and show that the response-time analysis developed for the protocol offers an upper bound on the response times.

Preliminary discussion on globally prioritized medium access for multi-channel wireless systems

We discuss the development of a simple globally prioritized multi-channel medium access control (MAC) protocol for wireless networks. This protocol provides “hard” pre-run-time real-time guarantees to sporadic message streams, exploits a very large fraction of the capacity of all channels for “hard” real-time traffic and also makes it possible to fully utilize the channels with non real-time traffic when hard real-time messages do not request to be transmitted. The potential of such protocols for real-time applications is discussed and a schedulability analysis is also presented.

Global static-priority preemptive multiprocessor scheduling with utilization bound 38%

Consider the problem of scheduling real-time tasks on a multiprocessor with the goal of meeting deadlines. Tasks arrive sporadically and have implicit deadlines, that is, the deadline of a task is equal to its minimum inter-arrival time. Consider this problem to be solved with global static-priority scheduling. We present a priority-assignment scheme with the property that if at most 38% of the processing capacity is requested then all deadlines are met.