No description available.

RETTES, Julio Alberto Sibaja. Robust algorithms for linear regression and locally linear embedding. 2017. 105 f. Dissertação (Mestrado em Ciência da Computação)- Universidade Federal do Ceará, Fortaleza, 2017. / Submitted by Weslayne Nunes de Sales (weslaynesales@ufc.br) on 2017-03-30T13:15:27Z No. of bitstreams: 1 2017_dis_rettesjas.pdf: 3569500 bytes, checksum: 46cedc2d9f96d0f58bcdfe3e0d975d78 (MD5) / Approved for entry into archive by Rocilda Sales (rocilda@ufc.br) on 2017-04-04T11:10:44Z (GMT) No. of bitstreams: 1 2017_dis_rettesjas.pdf: 3569500 bytes, checksum: 46cedc2d9f96d0f58bcdfe3e0d975d78 (MD5) / Made available in DSpace on 2017-04-04T11:10:44Z (GMT). No. of bitstreams: 1 2017_dis_rettesjas.pdf: 3569500 bytes, checksum: 46cedc2d9f96d0f58bcdfe3e0d975d78 (MD5) Previous issue date: 2017 / Nowadays a very large quantity of data is flowing around our digital society. There is a growing interest in converting this large amount of data into valuable and useful information. Machine learning plays an essential role in the transformation of data into knowledge. However, the probability of outliers inside the data is too high to marginalize the importance of robust algorithms. To understand that, various models of outliers are studied. In this work, several robust estimators within the generalized linear model for regression framework are discussed and analyzed: namely, the M-Estimator, the S-Estimator, the MM-Estimator, the RANSAC and the Theil-Sen estimator. This choice is motivated by the necessity of examining algorithms with different working principles. In particular, the M-, S-, MM-Estimator are based on a modification of the least squares criterion, whereas the RANSAC is based on finding the smallest subset of points that guarantees a predefined model accuracy. The Theil Sen, on the other hand, uses the median of least square models to estimate. The performance of the estimators under a wide range of experimental conditions is compared and analyzed. In addition to the linear regression problem, the dimensionality reduction problem is considered. More specifically, the locally linear embedding, the principal component analysis and some robust approaches of them are treated. Motivated by giving some robustness to the LLE algorithm, the RALLE algorithm is proposed. Its main idea is to use different sizes of neighborhoods to construct the weights of the points; to achieve this, the RAPCA is executed in each set of neighbors and the risky points are discarded from the corresponding neighborhood. The performance of the LLE, the RLLE and the RALLE over some datasets is evaluated. / Na atualidade um grande volume de dados é produzido na nossa sociedade digital. Existe um crescente interesse em converter esses dados em informação útil e o aprendizado de máquinas tem um papel central nessa transformação de dados em conhecimento. Por outro lado, a probabilidade dos dados conterem outliers é muito alta para ignorar a importância dos algoritmos robustos. Para se familiarizar com isso, são estudados vários modelos de outliers. Neste trabalho, discutimos e analisamos vários estimadores robustos dentro do contexto dos modelos de regressão linear generalizados: são eles o M-Estimator, o S-Estimator, o MM-Estimator, o RANSAC e o Theil-Senestimator. A escolha dos estimadores é motivada pelo principio de explorar algoritmos com distintos conceitos de funcionamento. Em particular os estimadores M, S e MM são baseados na modificação do critério de minimização dos mínimos quadrados, enquanto que o RANSAC se fundamenta em achar o menor subconjunto que permita garantir uma acurácia predefinida ao modelo. Por outro lado o Theil-Sen usa a mediana de modelos obtidos usando mínimos quadradosno processo de estimação. O desempenho dos estimadores em uma ampla gama de condições experimentais é comparado e analisado. Além do problema de regressão linear, considera-se o problema de redução da dimensionalidade. Especificamente, são tratados o Locally Linear Embedding, o Principal ComponentAnalysis e outras abordagens robustas destes. É proposto um método denominado RALLE com a motivação de prover de robustez ao algoritmo de LLE. A ideia principal é usar vizinhanças de tamanhos variáveis para construir os pesos dos pontos; para fazer isto possível, o RAPCA é executado em cada grupo de vizinhos e os pontos sob risco são descartados da vizinhança correspondente. É feita uma avaliação do desempenho do LLE, do RLLE e do RALLE sobre algumas bases de dados.

abstract: The power system is the largest man-made physical network in the world. Performing analysis of a large bulk system is computationally complex, especially when the study involves engineering, economic and environmental considerations. For instance, running a unit-commitment (UC) over a large system involves a huge number of constraints and integer variables. One way to reduce the computational expense is to perform the analysis on a small equivalent (reduced) model instead on the original (full) model. The research reported here focuses on improving the network reduction methods so that the calculated results obtained from the reduced model better approximate the performance of the original model. An optimization-based Ward reduction (OP-Ward) and two new generator placement methods in network reduction are introduced and numerical test results on large systems provide proof of concept. In addition to dc-type reductions (ignoring reactive power, resistance elements in the network, etc.), the new methods applicable to ac domain are introduced. For conventional reduction methods (Ward-type methods, REI-type methods), eliminating external generator buses (PV buses) is a tough problem, because it is difficult to accurately approximate the external reactive support in the reduced model. Recently, the holomorphic embedding (HE) based load-flow method (HELM) was proposed, which theoretically guarantees convergence given that the power flow equations are structure in accordance with Stahl’s theory requirements. In this work, a holomorphic embedding based network reduction (HE reduction) method is proposed which takes advantage of the HELM technique. Test results shows that the HE reduction method can approximate the original system performance very accurately even when the operating condition changes. / Dissertation/Thesis / Doctoral Dissertation Electrical Engineering 2017

The reliable construction of VLSI integrated circuits (ICs) requires that the ICs be tested after fabrication. An alternative to performing external testing is to create ICs that can test themselves with a built-in self-test (BIST) mode. Unfortunately the problem of embedding a self-test operating mode to the functional design is difficult for two reasons. (1) The creation of test sets that effectively test digital circuits requires the solution of several intractable problems. (2) The hardware resources dedicated to self-test are usually constrained. Modifications to the Logic III hardware description language and a new computer-aided design (CAD) tool, 1g3, are presented in this dissertation as an environment that allows BIST embedding to be created and evaluated. The major premise behind this work is that BIST design can be treated in a similar fashion as functional design, and that the designer can address the constraints of a BIST mode at the same time as the functional constraints. The modified language, called Logic III(UVic), allows BIST embeddings to be specified by an embedding module which describes how the circuit's memory elements are realized. This dissertation presents a library of embedding modules that realize several of the most common BIST architectures. Case studies using this environment are presented for an ALU, CORDIC, GCD, and string matching circuits. A BIST mode with almost 100% single stuck-at fault coverage is realized for each circuit. This shows that the CAD environment can be used to create self-testing circuits. In addition to aiding users in embedding BIST functionality, the 1g3 tool can be used to evaluate specific BIST architectures. Properties of BIST test pattern generators are presented that are used in analyzing the effectiveness of the generators for delay-fault testing. A novel approach based on creating a deterministic finite automaton that recognizes the fault-free responses is presented. / Graduate

Abstract Raw data sets taken with various capturing devices are usually multidimensional and need to be preprocessed before applying subsequent operations, such as clustering, classification, outlier detection, noise filtering etc. One of the steps of data preprocessing is dimensionality reduction. It has been developed with an aim to reduce or eliminate information bearing secondary importance, and retain or highlight meaningful information while reducing the dimensionality of data. Since the nature of real-world data is often nonlinear, linear dimensionality reduction techniques, such as principal component analysis (PCA), fail to preserve a structure and relationships in a highdimensional space when data are mapped into a low-dimensional space. This means that nonlinear dimensionality reduction methods are in demand in this case. Among them is a method called locally linear embedding (LLE), which is the focus of this thesis. Its main attractive characteristics are few free parameters to be set and a non-iterative solution avoiding the convergence to a local minimum. In this thesis, several extensions to the conventional LLE are proposed, which aid us to overcome some limitations of the algorithm. The study presents a comparison between LLE and three nonlinear dimensionality reduction techniques (isometric feature mapping (Isomap), self-organizing map (SOM) and fast manifold learning based on Riemannian normal coordinates (S-LogMap) applied to manifold learning. This comparison is of interest, since all of the listed methods reduce high-dimensional data in different ways, and it is worth knowing for which case a particular method outperforms others. A number of applications of dimensionality reduction techniques exist in data mining. One of them is visualization of high-dimensional data sets. The main goal of data visualization is to find a one, two or three-dimensional descriptive data projection, which captures and highlights important knowledge about data while eliminating the information loss. This process helps people to explore and understand the data structure that facilitates the choice of a proper method for the data analysis, e.g., selecting simple or complex classifier etc. The application of LLE for visualization is described in this research. The benefits of dimensionality reduction are commonly used in obtaining compact data representation before applying a classifier. In this case, the main goal is to obtain a low-dimensional data representation, which possesses good class separability. For this purpose, a supervised variant of LLE (SLLE) is proposed in this thesis.

In many instances the properties of a network are obstructed by an intervening network, which is required when performing measurements of the network. These intervening networks are often in the form of a mode transformer and are, in the general sense, referred to as error networks. A new analysis mechanism is developed by applying a de-embedding method that was identified as being robust. The analysis was subsequently implemented in a numerical computational software package. The analysis mechanism can then be applied to perform the characterisation of error networks. The performance of the analysis mechanism is verified using an ideal lumped-element network. The limitations of the mechanism are identified and possible ways of addressing these limitations are given. The mechanism is successfully applied to the characterisation of three different microwave networks. / Dissertation (MEng)--University of Pretoria, 2013. / gm2014 / Electrical, Electronic and Computer Engineering / unrestricted

Graph embedding, aiming to learn low-dimensional representations for nodes in graphs, has attracted increasing attention due to its critical application including node classification, link prediction and clustering in social network analysis. Most existing algorithms for graph embedding only rely on the topology information and fail to use the copious information in nodes as well as edges. As a result, their performance for many tasks may not be satisfactory. In this thesis, we proposed a novel and general framework for graph embedding with rich text information (GERI) through constructing a heterogeneous network, in which we integrate node and edge content information with graph topology. Specially, we designed a novel biased random walk to explore the constructed heterogeneous network with the notion of flexible neighborhood. Our sampling strategy can compromise between BFS and DFS local search on heterogeneous graph. To further improve our algorithm, we proposed semi-supervised GERI (SGERI), which learns graph embedding in an discriminative manner through heterogeneous network with label information. The efficacy of our method is demonstrated by extensive comparison experiments with 9 baselines over multi-label and multi-class classification on various datasets including Citeseer, Cora, DBLP and Wiki. It shows that GERI improves the Micro-F1 and Macro-F1 of node classification up to 10%, and SGERI improves GERI by 5% in Wiki.

There are two commonly discrete approximations for the inverse conductivity problem. Finite element models are heavily used in electrical impedance tomography research as they are easily adapted to bodies of irregular shapes. The other approximation is to use electrical resistor networks for which several uniqueness results and reconstruction algorithms are known for the inverse problem. In this thesis the link between finite element models and resistor networks is established. For the planar case we show how resistor networks associated with a triangular mesh have an isotropic embedding and we give conditions for the uniqueness of the embedding. Moreover, a layered finite element model parameterized by thevalues of conductivity on the interior nodes is constructed. Construction of the finite element mesh leads to a study of the triangulation survey problem. A constructive algorithm is given to determine the position of the nodes in the triangulation with a knowledge of one edge and the angles of the finite element mesh. Also we show that we need to satisfy the sine rule as aconsistency condition for every closed basic cycle that enclosing interior nodes and this is a complete set of independent constraints.

Dois problemas combinatórios são estudados: (i) determinar a quantidade de cópias de um hipergrafo fixo em um hipergrafo uniforme pseudoaleatório, e (ii) estimar números de Ramsey de ordem dois e três para grafos com largura de banda pequena e grau máximo limitado. Apresentamos um lema de contagem para estimar a quantidade de cópias de um hipergrafo k-uniforme linear livre de conectores (conector é uma generalização de triângulo, para hipergrafos) que estão presentes em um hipergrafo esparso pseudoaleatório G. Considere um hipergrafo k-uniforme linear H que é livre de conectores e um hipergrafo k-uniforme G com n vértices. Seja d_H=\\max\\{\\delta(J): J\\subset H\\} e D_H=\\min\\{k d_H,\\Delta(H)\\}. Estabelecemos que, se os vértices de G não possuem grau grande, famílias pequenas de conjuntos de k-1 elementos de V(G) não possuem vizinhança comum grande, e a maioria dos pares de conjuntos em {V(G)\\choose k-1} possuem a quantidade ``correta\'\' de vizinhos, então a quantidade de imersões de H em G é (1+o(1))n^{|V(H)|}p^{|E(H)|}, desde que p\\gg n^{1/D_H} e |E(G)|={n\\choose k}p. Isso generaliza um resultado de Kohayakawa, Rödl e Sissokho [Embedding graphs with bounded degree in sparse pseudo\\-random graphs, Israel J. Math. 139 (2004), 93--137], que provaram que, para p dado como acima, esse lema de imersão vale para grafos, onde H é um grafo livre de triângulos. Determinamos assintoticamente os números de Ramsey de ordem dois e três para grafos bipartidos com largura de banda pequena e grau máximo limitado. Mais especificamente, determinamos assintoticamente o número de Ramsey de ordem dois para grafos bipartidos com largura de banda pequena e grau máximo limitado, e o número de Ramsey de ordem três para tais grafos, com a suposição adicional de que ambas as classes do grafo bipartido têm aproximadamente o mesmo tamanho. / Two combinatorial problems are studied: (i) determining the number of copies of a fixed hipergraph in uniform pseudorandom hypergraphs, and (ii) estimating the two and three color Ramsey numbers for graphs with small bandwidth and bounded maximum degree. We give a counting lemma for the number of copies of linear k-uniform \\emph hypergraphs (connector is a generalization of triangle for hypergraphs) that are contained in some sparse hypergraphs G. Let H be a linear k-uniform connector-free hypergraph and let G be a k-uniform hypergraph with n vertices. Set d_H=\\max\\{\\delta(J)\\colon J\\subset H\\} and D_H=\\min\\{kd_H,\\Delta(H)\\}. We proved that if the vertices of G do not have large degree, small families of (k-1)-element sets of V(G) do not have large common neighbourhood and most of the pairs of sets in {V(G)\\choose k-1} have the `right\' number of common neighbours, then the number of embeddings of H in G is (1+o(1))n^p^, given that p\\gg n^ and |E(G)|=p. This generalizes a result by Kohayakawa, R\\\"odl and Sissokho [Embedding graphs with bounded degree in sparse pseudo\\-random graphs, Israel J. Math. 139 (2004), 93--137], who proved that, for p as above, this result holds for graphs, where H is a triangle-free graph. We determine asymptotically the two and three Ramsey numbers for bipartite graphs with small bandwidth and bounded maximum degree. More generally, we determine asymptotically the two color Ramsey number for bipartite graphs with small bandwidth and bounded maximum degree and the three color Ramsey number for such graphs with the additional assumption that both classes of the bipartite graph have almost the same size.

Tidigare forskning indikerar att organisationskultur är en viktig aspekt för organisatorisk framgång. Hur arbetet mot en gemensam organisationskultur i praktiken sker finns däremot motstridiga åsikter kring. Att sträva efter en väletablerad och stark organisationskultur kan försvåras av ett flertal aspekter, varav en decentraliserad organisationsstruktur är ett exempel. Syftet med undersökningen var att identifiera praktiska tillvägagångssätt för att stödja och förstärka en enhetlig organisationskultur i en verksamhet präglad av decentralisering. Undersökningen avsåg testa och vidareutveckla Scheins tolv mekanismer inom kulturskapande. En fallstudie genomfördes på Handelsbanken där affärsrådgivare, kontorschefer, rörelsechef, regionbankschef och delar av koncernledningen, inklusive personalchef, kreditchef och VD, intervjuats. Resultatet har visat att Handelsbankens tillvägagångssätt går att kategorisera in i Scheins modell för kulturskapande, men att beskrivningen av primära och sekundära faktorer inte överensstämmer med teorin. Istället utgör den decentraliserade strukturen en utgångspunkt för kulturen, där många övriga faktorer används i syfte att stödja decentraliseringen.