Supervised Aggregative Feature Extraction for Big Data Time Series Regression

In many applications, and especially those where batch processes are involved, a target scalar output of interest is often dependent on one or more time series of data. With the exponential growth in data logging in modern industries such time series are increasingly available for statistical modeling in soft sensing applications. In order to exploit time series data for predictive modelling, it is necessary to summarise the information they contain as a set of features to use as model regressors. Typically this is done in an unsupervised fashion using simple techniques such as computing statistical moments, principal components or wavelet decompositions, often leading to significant information loss and hence suboptimal predictive models. In this paper, a functional learning paradigm is exploited in a supervised fashion to derive continuous, smooth estimates of time series data (yielding aggregated local information), while simultaneously estimating a continuous shape function yielding optimal predictions. The proposed Supervised Aggregative Feature Extraction (SAFE) methodology can be extended to support nonlinear predictive models by embedding the functional learning framework in a Reproducing Kernel Hilbert Spaces setting. SAFE has a number of attractive features including closed form solution and the ability to explicitly incorporate first and second order derivative information. Using simulation studies and a practical semiconductor manufacturing case study we highlight the strengths of the new methodology with respect to standard unsupervised feature extraction approaches.

A practical taxonomy of methods and literature for managing uncertain spatial data in geographic information systems

Perfect information is seldom available to man or machines due to uncertainties inherent in real world problems. Uncertainties in geographic information systems (GIS) stem from either vague/ambiguous or imprecise/inaccurate/incomplete information and it is necessary for GIS to develop tools and techniques to manage these uncertainties. There is a widespread agreement in the GIS community that although GIS has the potential to support a wide range of spatial data analysis problems, this potential is often hindered by the lack of consistency and uniformity. Uncertainties come in many shapes and forms, and processing uncertain spatial data requires a practical taxonomy to aid decision makers in choosing the most suitable data modeling and analysis method. In this paper, we: (1) review important developments in handling uncertainties when working with spatial data and GIS applications; (2) propose a taxonomy of models for dealing with uncertainties in GIS; and (3) identify current challenges and future research directions in spatial data analysis and GIS for managing uncertainties.

Investigation into DCT Feature Selection for Visual Lip-Based Biometric Authentication

This paper investigated using lip movements as a behavioural biometric for person authentication. The system was trained, evaluated and tested using the XM2VTS dataset, following the Lausanne Protocol configuration II. Features were selected from the DCT coefficients of the greyscale lip image. This paper investigated the number of DCT coefficients selected, the selection process, and static and dynamic feature combinations. Using a Gaussian Mixture Model - Universal Background Model framework an Equal Error Rate of 2.20% was achieved during evaluation and on an unseen test set a False Acceptance Rate of 1.7% and False Rejection Rate of 3.0% was achieved. This compares favourably with face authentication results on the same dataset whilst not being susceptible to spoofing attacks.

k-fold Subsampling based Sequential Backward Feature Elimination

We present a new wrapper feature selection algorithm for human detection. This algorithm is a hybrid featureselection approach combining the benefits of filter and wrapper methods. It allows the selection of an optimalfeature vector that well represents the shapes of the subjects in the images. In detail, the proposed featureselection algorithm adopts the k-fold subsampling and sequential backward elimination approach, while thestandard linear support vector machine (SVM) is used as the classifier for human detection. We apply theproposed algorithm to the publicly accessible INRIA and ETH pedestrian full image datasets with the PASCALVOC evaluation criteria. Compared to other state of the arts algorithms, our feature selection based approachcan improve the detection speed of the SVM classifier by over 50% with up to 2% better detection accuracy.Our algorithm also outperforms the equivalent systems introduced in the deformable part model approach witharound 9% improvement in the detection accuracy

Progress in the use of scanning electron microscopy (SEM) applied to the taxonomy of the genus Bursaphelenchus (Nematoda: Parasitaphelenchidae)

The scanning electron microscope (SEM) has been a major tool in detailed morphological observations of plant parasitic nematodes during the last 30 years, efficiently complementing light microscopical (LM) studies. Nematodes are extremely difficult to observe and characterize due to their small size (aprox. 1 mm long) and paucity of morphological characters, so detailed surface observations of several organs and nematode regions are of the highest value. Among plant parasitic nematodes, one of the most devastating species is the “pinewood nematode” (PWN), Bursaphelenchus xylophilus, which has been a major problem for forest species, and in particular pines, in Asia (Japan, China, Korea) and has been recently detected in the European Union (Portugal). B. xylophilus belongs to a closely related, morphologically similar group of species, within the genus Bursaphelenchus, and designated by the “xylophilus group”. SEM has become a crucial tool in observing several genital characters of males and females, such as male genital papillae, male copulatory spicules, female vulval flap and female genital papillae.s In this presentation, we will show how SEM has been utilized to observe and characterize the shape of the vulval flap, the presence/ absence of papillae near the flap, and confirm the presence and the arrangement of the male genital papillae. LM is also used in this work to show its value as a complementary tool to SEM, in both genital characteristics and other, general, characters of the genus Bursaphelenchus, such as the male bursa and cephalic region.

Classification of incomplete feature vectors by radial basis function networks

The paper describes the use of radial basis function neural networks with Gaussian basis functions to classify incomplete feature vectors. The method uses the fact that any marginal distribution of a Gaussian distribution can be determined from the mean vector and covariance matrix of the joint distribution.

Identification of descendants of an extinct bovine population from the Algarve region of Portugal using numerical taxonomy analysis of morphological traits

The morphology of a sample of four bulls and 43 cows, presumed to be descendants of the extinct cattle breed ‘Algarvia’ (AG), was used to assign their relationship with animals from other Portuguese autochthonous breeds – Arouquesa (AR), Barrosa˜ (BA), Cachena (CA), Marinhoa (MA), Maronesa (MO), Minhota (MN), Mirandesa (MI), (only bulls), Alentejana (AL), Garvonesa (GA), Mertolenga (ME) and Preta (PR). Standard numerical taxonomic methods were applied to a set of 183 (cows) and 170 (bulls) traits, to derive average pairwise taxonomic distances among the sample of 257 cows and 76 bulls. Distance coefficients (morphological index of distance) ranged from 0.22 to 2.62 (cows) and from 0.49 to 2.13 (bulls). Unweighted pair group method using arithmetic averages (UPGMA)-based phenograms and a principal coordinate analysis showed that bulls were highly clustered and cows showed a tendency to cluster according to their geographical and breed origin. The AG population grouped together with GA, AL, ME and MN breeds in the Red Convex group. The average taxonomic distance among breeds was 1.02, the highest being 1.39 (ME versus BA) and the lowest being 0.64 (MA versus AR). The approach allowed for the identification of a phenotypically differentiated set of animals, comprising 19 cows and four bulls representative of the AG breed, and which can be targeted in further studies aiming at the recovery of this extinct breed.

Microalgae taxonomy and breeding

Microalgae are a diverse group of organisms that form the basic component of many eco­- systems. The systemic classification of algae is primarily based on their pigment composition and is divided into nine classes. The largest groups are Chlorophyceae (green algae), Phaeophyceae (brown algae), Pyrrophyceae (dinoflagellates), Chrysophyceae (golden brown algae), Bacillariophyceae (diatoms) and Rhodo­ phyceae (red algae). It has been estimated that between 22,000 and 26,000 species exist (Norton et al., 1996), of which only a few species have been identified to be useful for commercial application, such as Spirulina, Chlorella, Haematococcus, Dunaliella, Botryococcus, Phaeodactylum and Porphyridium. Several other species that are also cultivated commercially for the hatcheries in the aquaculture field include Chaetoceros, Crypthecodinium, Isochrysis, Nannochloris, Nitzschia, Schizo­chytrium, Tetraselmis and Skeletonema (Raja et al., 2008).

A Taxonomy for Developing Cultural Competence

This paper proposes a taxonomy to develop culturally competent practitioners. Arguments about what this might mean and how this could be achieved are discussed first, identifying problems with multicultural and antiracist approaches. The model follows the cognitive, emotional and behavioural levels of Steinaker and Bell's experiential taxonomy. Five elements are proposed: cultural awareness, cultural knowledge, cultural understanding, cultural sensitivity and cultural competence. These could address, in increasingly sophisticated and increasingly praxis-oriented ways, issues of power and the construction of meanings and identities which go beyond essentialist notions of ethnicity.

Feature Engineering in Fine-Grained Image Classification

Thesis (Ph.D.)--University of Washington, 2013

Taxonomy of fundamental concepts of localization in cyber-physical and sensor networks

Localization is a fundamental task in Cyber-Physical Systems (CPS), where data is tightly coupled with the environment and the location where it is generated. The research literature on localization has reached a critical mass, and several surveys have also emerged. This review paper contributes on the state-of-the-art with the proposal of a new and holistic taxonomy of the fundamental concepts of localization in CPS, based on a comprehensive analysis of previous research works and surveys. The main objective is to pave the way towards a deep understanding of the main localization techniques, and unify their descriptions. Furthermore, this review paper provides a complete overview on the most relevant localization and geolocation techniques. Also, we present the most important metrics for measuring the accuracy of localization approaches, which is meant to be the gap between the real location and its estimate. Finally, we present open issues and research challenges pertaining to localization. We believe that this review paper will represent an important and complete reference of localization techniques in CPS for researchers and practitioners and will provide them with an added value as compared to previous surveys.

Incremental filter and wrapper approaches for feature discretization

Discrete data representations are necessary, or at least convenient, in many machine learning problems. While feature selection (FS) techniques aim at finding relevant subsets of features, the goal of feature discretization (FD) is to find concise (quantized) data representations, adequate for the learning task at hand. In this paper, we propose two incremental methods for FD. The first method belongs to the filter family, in which the quality of the discretization is assessed by a (supervised or unsupervised) relevance criterion. The second method is a wrapper, where discretized features are assessed using a classifier. Both methods can be coupled with any static (unsupervised or supervised) discretization procedure and can be used to perform FS as pre-processing or post-processing stages. The proposed methods attain efficient representations suitable for binary and multi-class problems with different types of data, being competitive with existing methods. Moreover, using well-known FS methods with the features discretized by our techniques leads to better accuracy than with the features discretized by other methods or with the original features. (C) 2013 Elsevier B.V. All rights reserved.

A importância do capital humano nas organizações

Os ativos intangíveis são cada vez mais uma preocupação das organizações, e atualmente são reconhecidos como os principais ativos das empresas. O Capital Humano como dimensão do Capital Intelectual é um fator preponderante no desenvolvimento e crescimento das organizações, uma vez que proporciona criação de valor e vantagem competitiva para as empresas. A criação, a partilha e a transferência de Conhecimento são, também, fatores influentes que geram Capital Humano. Na atualidade, este tema tem despertado o interesse tanto de economistas, gestores e contabilistas, como de meros investidores. O Capital Intelectual é tradicionalmente concetualizado como sendo composto por três grandes dimensões: Capital Humano, Capital Relacional e Capital Estrutural. Por sua vez, daquilo que é o nosso conhecimento, consideramos que existe uma lacuna na literatura sobre Capital Humano no que diz respeito à sua taxonomia. Efetivamente grande parte das investigações sobre Capital Humano, como dimensão do Capital Intelectual, focam-se essencialmente nos itens necessários para a sua mensuração do CH. Desta forma, o objetivo principal deste estudo consiste em explorar a dimensão do CH ao nível das suas componentes. Ou seja, procuramos encontrar as componentes do Capital Humano e propomo-nos a determinar quais as que têm maior importância no CH para o desenvolvimento e crescimento das organizações. Para esta realização efetuámos um estudo de carácter exploratório, num contexto específico do mercado português – o Setor Segurador. Os resultados obtidos tanto a nível qualitativo como quantitativo vão de encontro às questões de investigação previamente definidas. Portanto, as componentes mais importantes do CH são: a formação e o bem-estar, o conhecimento e o profissionalismo e as características pessoais e técnicas dos colaboradores que constituem os Recursos Humanos da organização. Estas são aquelas que mais valorizam e proporcionam crescimento nas organizações. Este estudo poderia ser tão mais completo, se pudéssemos apresentar uma comparação entre duas empresas seguradoras e, consequentemente analisar o comportamento das duas face ao Capital Humano. Outro aspeto interessante seria efetuar uma análise sobre qual o impacto do Capital Humano na performance financeira das organizações seguradoras. Estas são limitações que podem ser vistas como sugestões para estudos de investigação futuros nesta mesma área. Este estudo contribui para o enriquecimento das investigações na área do Capital Humano, uma vez que conhecendo melhor as componentes que constituem o CH mais facilmente as organizações definem as suas estratégicas de crescimento e desenvolvimento. Desta forma, este estudo pode apoiar alguns gestores na definição de políticas de valorização deste ativo intangível em organizações do mercado segurador.