Studying Recommender Systems to Enhance Distributed Computing Schedulers

Distributed Computing frameworks belong to a class of programming models that allow developers to

launch workloads on large clusters of machines. Due to the dramatic increase in the volume of

data gathered by ubiquitous computing devices, data analytic workloads have become a common

case among distributed computing applications, making Data Science an entire field of

Computer Science. We argue that Data Scientist's concern lays in three main components: a dataset,

a sequence of operations they wish to apply on this dataset, and some constraint they may have

related to their work (performances, QoS, budget, etc). However, it is actually extremely

difficult, without domain expertise, to perform data science. One need to select the right amount

and type of resources, pick up a framework, and configure it. Also, users are often running their

application in shared environments, ruled by schedulers expecting them to specify precisely their resource

needs. Inherent to the distributed and concurrent nature of the cited frameworks, monitoring and

profiling are hard, high dimensional problems that block users from making the right

configuration choices and determining the right amount of resources they need. Paradoxically, the

system is gathering a large amount of monitoring data at runtime, which remains unused.

In the ideal abstraction we envision for data scientists, the system is adaptive, able to exploit

monitoring data to learn about workloads, and process user requests into a tailored execution

context. In this work, we study different techniques that have been used to make steps toward

such system awareness, and explore a new way to do so by implementing machine learning

techniques to recommend a specific subset of system configurations for Apache Spark applications.

Furthermore, we present an in depth study of Apache Spark executors configuration, which highlight

the complexity in choosing the best one for a given workload.

Algorithms for Geometric Matching, Clustering, and Covering

With the popularization of GPS-enabled devices such as mobile phones, location data are becoming available at an unprecedented scale. The locations may be collected from many different sources such as vehicles moving around a city, user check-ins in social networks, and geo-tagged micro-blogging photos or messages. Besides the longitude and latitude, each location record may also have a timestamp and additional information such as the name of the location. Time-ordered sequences of these locations form trajectories, which together contain useful high-level information about people's movement patterns.

The first part of this thesis focuses on a few geometric problems motivated by the matching and clustering of trajectories. We first give a new algorithm for computing a matching between a pair of curves under existing models such as dynamic time warping (DTW). The algorithm is more efficient than standard dynamic programming algorithms both theoretically and practically. We then propose a new matching model for trajectories that avoids the drawbacks of existing models. For trajectory clustering, we present an algorithm that computes clusters of subtrajectories, which correspond to common movement patterns. We also consider trajectories of check-ins, and propose a statistical generative model, which identifies check-in clusters as well as the transition patterns between the clusters.

The second part of the thesis considers the problem of covering shortest paths in a road network, motivated by an EV charging station placement problem. More specifically, a subset of vertices in the road network are selected to place charging stations so that every shortest path contains enough charging stations and can be traveled by an EV without draining the battery. We first introduce a general technique for the geometric set cover problem. This technique leads to near-linear-time approximation algorithms, which are the state-of-the-art algorithms for this problem in either running time or approximation ratio. We then use this technique to develop a near-linear-time algorithm for this

shortest-path cover problem.

Current meter measurements in the equatorial Atlantic

From 27 January to 23 June 1979 R.V. "Meteor" surveyed the central equatorial Atlantic on a section along 22° W from 3° N to 2° S. During the observation period, a hydrographic section down to 600 m was repeated ten times with a continous "Howaldt-Bathysonde" CTD and a rosette sampler. The station distance was 10 to 15 nm. The water samples were used to recalibrate salinity and to determine oxygen, nutrients and chlorophyll a. An undulating CTD system ("Delphin") was towed on 11 sections. A profiling distance of one to two nautical miles and a profile depth of 90 m was obtained. Five current meter arrys were moored along 22° W between 3° N and 1°S from January to March 1979. In May and June two moorings were installed at 2° N and at the equator. On the buoys measurements of wind speed and direction were obtained. At 43 stations a wire-guided Aanderaa profiling current meter was successfully lowered. Drifting buoy experiments were repeated three times with clusters of 5 to 10 buoys. A fourth experiment took place in 1978 in the Gulf of Guinea. On the way from and to XBT sections were carried out. The data sets obtained by these instruments are presented in this data report.

Seawater carbonate chemistry and Bathymodiolus brevior shell variables near Eifuku volcano, Japan, 2009

Increasing atmospheric carbon dioxide levels are causing ocean acidification, compromising the ability of some marine organisms to build and maintain support structures as the equilibrium state of inorganic carbon moves away from calcium carbonate. Few marine organisms tolerate conditions where ocean pH falls significantly below today's value of about 8.1 and aragonite and calcite saturation values below 1. Here we report dense clusters of the vent mussel B. brevior in natural conditions of pH values between 5.36 and 7.29 on northwest Eifuku volcano, Mariana arc, where liquid carbon dioxide and hydrogen sulphide emerge in a hydrothermal setting. We find that both shell thickness and daily growth increments in shells from northwest Eifuku are only about half those recorded from mussels living in water with pH>7.8. Low pH may therefore also be implicated in metabolic impairment. We identify four-decade-old mussels, but suggest that the mussels can survive for so long only if their protective shell covering remains intact: crabs that could expose the underlying calcium carbonate to dissolution are absent from this setting. The mussels' ability to precipitate shells in such low-pH conditions is remarkable. Nevertheless, the vulnerability of molluscs to predators is likely to increase in a future ocean with low pH.

Feature Selection for Anomaly Detection Using Optical Emission Spectroscopy

To maintain the pace of development set by Moore's law, production processes in semiconductor manufacturing are becoming more and more complex. The development of efficient and interpretable anomaly detection systems is fundamental to keeping production costs low. As the dimension of process monitoring data can become extremely high anomaly detection systems are impacted by the curse of dimensionality, hence dimensionality reduction plays an important role. Classical dimensionality reduction approaches, such as Principal Component Analysis, generally involve transformations that seek to maximize the explained variance. In datasets with several clusters of correlated variables the contributions of isolated variables to explained variance may be insignificant, with the result that they may not be included in the reduced data representation. It is then not possible to detect an anomaly if it is only reflected in such isolated variables. In this paper we present a new dimensionality reduction technique that takes account of such isolated variables and demonstrate how it can be used to build an interpretable and robust anomaly detection system for Optical Emission Spectroscopy data.

Construção de um Referencial de Competências dos Técnicos de Análises Clínicas e Saúde Pública

O presente estudo teve como objectivo construir e validar um referencial de competências dos Técnicos de Análises Clínicas e Saúde Pública. Para a concretização deste objectivo construi-se uma entrevista estruturada com 15 questões de resposta aberta aplicada a uma amostra de 10 Técnicos de Análises Clínicas e Saúde Pública. Os dados foram recolhidos, culminaram num questionário de resposta fechada constituído por 36 competências e aplicado a 100 técnicos. Dos resultados do tratamento deste, derivou um outro questionário com 26 competências, relativamente às quais, além da classificação das mesmas, era solicitado emissão de opinião técnica. Este instrumento foi aplicado a 10 peritos (professores que ministraram ou ministram na Licenciatura de Análises Clínicas e Saúde Pública e/ou investigadores desta área de estudos). Daí resultaram 22 competências que foram sujeitas a validação semântica e que constituíram o referencial de competências. O conjunto que competências a que se chegou foi trabalhado em clusters de competências pessoais, interpessoais e instrumentais e para se verificar a sua fiabilidade interna aplicou-se o Alfa de Cronbach. O referencial de competências constituído foi validado através de um Focus Group de técnicos. A construção do referencial de competências dos Técnicos de Análises Clínicas e Saúde Pública permitiu dar a conhecer as competências essenciais à apresentação de um bom desempenho profissional, quando sujeitos a avaliação por competências, e o desenvolvimento de estudos acerca desta profissão. / The following paper aimed to develop and validate a reference of the Clinical Analysis and Public Health technicians’ competences. In order to achieve this purpose it was put together a structured interview based on fifteen open-response questions. This interview was applied to a sample of ten Clinical Analysis and Public Health technicians. The collected data of this interview led to a closed response questionnaire including thirty-six competences, which was then administered to a hundred technicians. Subsequently, the data collected in this questionnaire let to another different questionnaire composed of twenty-six competences. This time, besides their classification it was required to issue a technical opinion. This research instrument was then administered to ten experts (teachers who minister or ministered the Degree of Clinical Analysis and Public Health and / or researchers in this field of study). This last resulted into twenty-two competences which were subjected to semantic validation and afterwards composed the reference of competences. The reached reference of competences was then worked on to clusters of personal, interpersonal and instrumental competences whose liability was checked through the Cronbach's Alpha. The reference of competences was then validated through a technician’s Focus Group. The building of this Clinical Analysis and Public Health technicians’ reference of competences was important to acknowledge the important competences necessary to perform this Job well and professionally.

Adsorption Calorimetry for Energy Conversion Technologies: Applications in Organic Photovoltaics, Catalysis, and Atomic Layer Deposition

Thesis (Ph.D.)--University of Washington, 2016-07

Mesure des comportements des mentors en contexte de mentorat scolaire et liens entre ces comportements, la qualité de la relation de mentorat et l’ajustement des protégés

Les résultats d’études récentes suggèrent que certains comportements de soutien des mentors pourraient augmenter les bénéfices du mentorat scolaire. Cependant, peu d’outils validés sont disponibles dans la littérature pour mesurer ces comportements. Il est aussi convenu que l’efficacité du mentorat scolaire dépend de la qualité de la relation tissée entre le mentor et son protégé, laquelle serait tributaire d’un ensemble de facteurs, dont les comportements du mentor. Néanmoins, encore une fois très peu d’études empiriques ont tenté d’identifier les patrons de comportements des mentors les plus susceptibles d’influencer la relation de mentorat et l’ajustement des protégés. La présente thèse poursuit deux objectifs, soit de construire et valider un outil de mesure des comportements de soutien des mentors oeuvrant en contexte de mentorat scolaire, puis d’explorer les liens entre des comportements de structure et de soutien des mentors, la relation de mentorat et l’ajustement des protégés. L’échelle de comportements des mentors (ECM) a été développée en s’inspirant des prémisses du modèle sociomotivationnel du mentorat (Larose & Tarabulsy, 2014). Deux cent cinquante-trois étudiants du collégial participant à un programme de mentorat scolaire d’une durée de huit mois ont complété une version expérimentale de l’ECM ainsi que différentes mesures de la qualité de la relation de mentorat à deux temps de leur participation au programme. Les résultats montrent que le questionnaire possède de bons coefficients de cohérence interne et une structure factorielle adéquate, à l’exception du facteur soutien à l’autonomie. De plus, trois des dimensions de l’ECM prédisent la qualité de la relation de mentorat et la perception d’utilité de l’intervention. Des recommandations pour l’utilisation et l’amélioration de l’ECM sont proposées. Sur la base des évaluations des protégés à l’ECM (Brodeur et al., 2015), quatre regroupements distincts de comportements de mentors ont été identifiés : Optimal, Suffisant, Contrôlant, et Inadéquat. Les résultats montrent que plus les mentors ont fait preuve de soutien et de structure, plus les protégés ont évalué positivement la relation et l’utilité du mentorat, sauf pour les profils Optimal et Contrôlant. Par ailleurs, uniquement l’ajustement social des protégés a différé du groupe contrôle, et ce proportionnellement à la quantité de soutien et de structure prodigués par les mentors. D’autre part, il est discuté de l’impact de l’ajustement initial des protégés sur les comportements des mentors. Des implications théoriques et pratiques des résultats des deux articles sont présentées. Mots-clés : mentorat scolaire, comportements des mentors, validation de questionnaire, qualité de la relation de mentorat, ajustement des protégés.

A fuzzy clustering approach to a demand response model

This paper proposes a novel demand response model using a fuzzy subtractive cluster approach. The model development provides support to domestic consumer decisions on controllable loads management, considering consumers' consumption needs and the appropriate load shape or rescheduling in order to achieve possible economic benefits. The model based on fuzzy subtractive clustering method considers clusters of domestic consumption covering an adequate consumption range. Analysis of different scenarios is presented considering available electric power and electric energy prices. Simulation results are presented and conclusions of the proposed demand response model are discussed. (C) 2016 Elsevier Ltd. All rights reserved.

Território de Urbanização Difusa

A presente dissertação constitui o culminar do Mestrado Integrado em Arquitetura e Urbanismo, da Universidade Fernando Pessoa – Porto. Tem como objeto de estudo a indústria em Território de Urbanização Difusa, sendo o objetivo maior a sua compreensão através, principalmente, da relação entre a dispersão industrial e as diretrizes de ocupação determinadas pelos planos de ordenamento do território, no contexto do concelho. Com este intuito foram analisadas quatro zonas que abrangem parte das freguesias de Gandra, Rebordosa, Lordelo, Vilela, Sobrosa e Duas Igrejas, localizadas no noroeste do concelho de Paredes, as quais se observaram num período cronológico delimitado entre 1947 e 2011. Trata-se de um território de grande complexidade onde existe uma multifuncionalidade de espaços. Caracteriza-se, fundamentalmente, por uma ocupação difusa/dispersa, tendo como suporte a rede de estradas e caminhos públicos em que a atividade industrial predomina nas mais diversas dimensões, desde as zonas industriais aquela inserida nos aglomerados residenciais de cariz familiar. Esta dispersão intensificou-se ao longo do séc. XX, com a fixação dispersa das indústrias que promoveram o desenvolvimento de um modelo de economia familiar que se reflete no território. A implementação do Planeamento formal, em finais desse século, virá estabelecer regras que ainda assim parecem insuficientes para controlar os aspetos mais negativos das opções individuais pré-plano. Na verdade, o atual modelo de implantação industrial levanta muitas preocupações no que concerne à preservação das estruturas “verdes”, podendo vir a comprometer, no futuro, a sustentabilidade ambiental e ecológica deste território. Será com algumas breves reflexões a este propósito, após repassar as questões principais do trabalho, que o terminaremos.

Clustering through post inhibitory rebound in synaptically coupled neurons

Post inhibitory rebound is a nonlinear phenomenon present in a variety of nerve cells. Following a period of hyper-polarization this effect allows a neuron to fire a spike or packet of spikes before returning to rest. It is an important mechanism underlying central pattern generation for heartbeat, swimming and other motor patterns in many neuronal systems. In this paper we consider how networks of neurons, which do not intrinsically oscillate, may make use of inhibitory synaptic connections to generate large scale coherent rhythms in the form of cluster states. We distinguish between two cases i) where the rebound mechanism is due to anode break excitation and ii) where rebound is due to a slow T-type calcium current. In the former case we use a geometric analysis of a McKean type model to obtain expressions for the number of clusters in terms of the speed and strength of synaptic coupling. Results are found to be in good qualitative agreement with numerical simulations of the more detailed Hodgkin-Huxley model. In the second case we consider a particular firing rate model of a neuron with a slow calcium current that admits to an exact analysis. Once again existence regions for cluster states are explicitly calculated. Both mechanisms are shown to prefer globally synchronous states for slow synapses as long as the strength of coupling is sufficiently large. With a decrease in the duration of synaptic inhibition both systems are found to break into clusters. A major difference between the two mechanisms for cluster generation is that anode break excitation can support clusters with several groups, whilst slow T-type calcium currents predominantly give rise to clusters of just two (anti-synchronous) populations.

A Pyramidal Genetic Algorithm for Multiple-Choice Problems

This paper combines the idea of a hierarchical distributed genetic algorithm with different inter-agent partnering strategies. Cascading clusters of sub-populations are built from bottom up, with higher-level sub-populations optimising larger parts of the problem. Hence, higher-level sub-populations search a larger search space with a lower resolution whilst lower-level sub-populations search a smaller search space with a higher resolution. The effects of different partner selection schemes amongst the agents on solution quality are examined for two multiple-choice optimisation problems. It is shown that partnering strategies that exploit problem-specific knowledge are superior and can counter inappropriate (sub-) fitness measurements.

Partnering Strategies for Fitness Evaluation in a Pyramidal Evolutionary Algorithm

This paper combines the idea of a hierarchical distributed genetic algorithm with different inter-agent partnering strategies. Cascading clusters of sub-populations are built from bottom up, with higher-level sub-populations optimising larger parts of the problem. Hence higher-level sub-populations search a larger search space with a lower resolution whilst lower-level sub-populations search a smaller search space with a higher resolution. The effects of different partner selection schemes for (sub-)fitness evaluation purposes are examined for two multiple-choice optimisation problems. It is shown that random partnering strategies perform best by providing better sampling and more diversity.

A Pyramidal Evolutionary Algorithm with Different Inter-Agent Partnering Strategies for Scheduling Problems

This paper combines the idea of a hierarchical distributed genetic algorithm with different inter-agent partnering strategies. Cascading clusters of sub-populations are built from bottom up, with higher-level sub-populations optimising larger parts of the problem. Hence higher-level sub-populations search a larger search space with a lower resolution whilst lower-level sub-populations search a smaller search space with a higher resolution. The effects of different partner selection schemes amongst the agents on solution quality are examined for two multiple-choice optimisation problems. It is shown that partnering strategies that exploit problem-specific knowledge are superior and can counter inappropriate (sub-) fitness measurements.

Dificuldades de aprendizagem específicas: respostas educativas diferenciadas

Dissertação apresentada à Escola Superior de Educação de Paula Frassinetti, para a obtenção do grau de mestre em Ciências da Educação. Especialização em Educação Especial