Evolutionary model trees for handling continuous classes in machine learning

Model trees are a particular case of decision trees employed to solve regression problems. They have the advantage of presenting an interpretable output, helping the end-user to get more confidence in the prediction and providing the basis for the end-user to have new insight about the data, confirming or rejecting hypotheses previously formed. Moreover, model trees present an acceptable level of predictive performance in comparison to most techniques used for solving regression problems. Since generating the optimal model tree is an NP-Complete problem, traditional model tree induction algorithms make use of a greedy top-down divide-and-conquer strategy, which may not converge to the global optimal solution. In this paper, we propose a novel algorithm based on the use of the evolutionary algorithms paradigm as an alternate heuristic to generate model trees in order to improve the convergence to globally near-optimal solutions. We call our new approach evolutionary model tree induction (E-Motion). We test its predictive performance using public UCI data sets, and we compare the results to traditional greedy regression/model trees induction algorithms, as well as to other evolutionary approaches. Results show that our method presents a good trade-off between predictive performance and model comprehensibility, which may be crucial in many machine learning applications. (C) 2010 Elsevier Inc. All rights reserved.

Aproximaciones del conjunto eficiente en decisión multicriterio

El conjunto eficiente en la Teoría de la Decisión Multicriterio juega un papel fundamental en los procesos de solución ya que es en este conjunto donde el decisor debe hacer su elección más preferida. Sin embargo, la generación de tal conjunto puede ser difícil, especialmente en problemas continuos y/o no lineales. El primer capítulo de esta memoria, es introductorio a la Decisión Multicriterio y en él se exponen aquellos conceptos y herramientas que se van a utilizar en desarrollos posteriores. El segundo capítulo estudia los problemas de Toma de Decisiones en ambiente de certidumbre. La herramienta básica y punto de partida es la función de valor vectorial que refleja imprecisión sobre las preferencias del decisor. Se propone una caracterización del conjunto de valor eficiente y diferentes aproximaciones con sus propiedades de encaje y convergencia. Varios algoritmos interactivos de solución complementan los desarrollos teóricos. El tercer capítulo está dedicado al caso de ambiente de incertidumbre. Tiene un desarrollo parcialmente paralelo al anterior y utiliza la función de utilidad vectorial como herramienta de modelización de preferencias del decisor. A partir de la consideración de las distribuciones simples se introduce la eficiencia en utilidad, su caracterización y aproximaciones, que posteriormente se extienden a los casos de distribuciones discretas y continuas. En el cuarto capítulo se estudia el problema en ambiente difuso, aunque de manera introductoria. Concluimos sugiriendo distintos problemas abiertos.---ABSTRACT---The efficient set of a Multicriteria Decicion-Making Problem plays a fundamental role in the solution process since the Decisión Maker's preferred choice should be in this set. However, the computation of that set may be difficult, specially in continuous and/or nonlinear problems. Chapter one introduces Multicriteria Decision-Making. We review basic concepts and tools for later developments. Chapter two studies Decision-Making problems under certainty. The basic tool is the vector valué function, which represents imprecisión in the DM's preferences. We propose a characterization of the valué efficient set and different approximations with nesting and convergence properties. Several interactive algorithms complement the theoretical results. We devote Chapter three to problems under uncertainty. The development is parallel to the former and uses vector utility functions to model the DM's preferences. We introduce utility efficiency for simple distributions, its characterization and some approximations, which we partially extend to discrete and continuous classes of distributions. Chapter four studies the problem under fuzziness, at an exploratory level. We conclude with several open problems.

Minimal Nielsen Root Classes and Roots of Liftings

Given a continuous map f : K -> M from a 2-dimensional CW complex into a closed surface, the Nielsen root number N(f) and the minimal number of roots mu(f) of f satisfy N(f) <= mu(f). But, there is a number mu(C)(f) associated to each Nielsen root class of f, and an important problem is to know when mu(f) = mu(C)(f)N(f). In addition to investigate this problem, we determine a relationship between mu(f) and mu((f) over tilde), when (f) over tilde f is a lifting of f through a covering space, and we find a connection between this problems, with which we answer several questions related to them when the range of the maps is the projective plane.

Continuous reproduction and recruitment in the narrowback mud crab Panopeus americanus (Brachyura, Panopeidae) in a remnant human-impacted mangrove area

The population structure, reproductive period, and juvenile recruitment of Panopeus americanus were studied in order to enhance knowledge of its life cycle and reproductive strategy and promote the maintenance of its natural stocks in an impacted region. Specimens were collected in the remnant human-impacted mangrove at Araca, state of Sao Paulo, Brazil, at two-month intervals from January to November 2005, at low tide, with a capture effort by three people. The crabs were measured (carapace width, CW) and sexed. The total catch was 275 animals, including 132 males (48.0%); 127 females (46.2%), of which 39 were ovigerous (14.2% of total catch); and 16 individuals whose sex could not be identified (5.8%). No correlation was observed between water temperature and the number of collected individuals; however, there was a significant, positive correlation with salinity. Males were significantly larger than females. The size-frequency distribution was bimodal, reflecting the occurrende of more than one recruitment pulse and the differential abundance of adults during the period of study. The overall sex ratio was 1:0.97 favoring males, and was not significantly different from the expected value, i.e., this population fits the anomalous pattern of sex occurrence in size classes. Ovigerous females were captured in all sampled months, which explain the continuous recruitment observed. Expected low levels of individuals of different size classes in the population were not confirmed. All population aspects found here allowed us to infer that this population of P. americanus is well established in the impacted mangrove by virtue of its reproductive strategy.

Towards input/output-free modelling of linear infinite-dimensional systems in continuous time

This dissertation describes a networking approach to infinite-dimensional systems theory, where there is a minimal distinction between inputs and outputs. We introduce and study two closely related classes of systems, namely the state/signal systems and the port-Hamiltonian systems, and describe how they relate to each other. Some basic theory for these two classes of systems and the interconnections of such systems is provided. The main emphasis lies on passive and conservative systems, and the theoretical concepts are illustrated using the example of a lossless transfer line. Much remains to be done in this field and we point to some directions for future studies as well.

The Individual is Nothing, the Class Everything: Psychophysics and Modeling of Recognition in Obect Classes

Most psychophysical studies of object recognition have focussed on the recognition and representation of individual objects subjects had previously explicitely been trained on. Correspondingly, modeling studies have often employed a 'grandmother'-type representation where the objects to be recognized were represented by individual units. However, objects in the natural world are commonly members of a class containing a number of visually similar objects, such as faces, for which physiology studies have provided support for a representation based on a sparse population code, which permits generalization from the learned exemplars to novel objects of that class. In this paper, we present results from psychophysical and modeling studies intended to investigate object recognition in natural ('continuous') object classes. In two experiments, subjects were trained to perform subordinate level discrimination in a continuous object class - images of computer-rendered cars - created using a 3D morphing system. By comparing the recognition performance of trained and untrained subjects we could estimate the effects of viewpoint-specific training and infer properties of the object class-specific representation learned as a result of training. We then compared the experimental findings to simulations, building on our recently presented HMAX model of object recognition in cortex, to investigate the computational properties of a population-based object class representation as outlined above. We find experimental evidence, supported by modeling results, that training builds a viewpoint- and class-specific representation that supplements a pre-existing repre-sentation with lower shape discriminability but possibly greater viewpoint invariance.

On isomorphism classes of C(2(m) circle plus [0, alpha]) spaces

We provide a complete isomorphic classification of the Banach spaces of continuous functions on the compact spaces 2(m) circle plus [0, alpha], the topological sums of Cantor cubes 2(m), with m smaller than the first sequential cardinal, and intervals of ordinal numbers [0, alpha]. In particular, we prove that it is relatively consistent with ZFC that the only isomorphism classes of C(2(m) circle plus [0, alpha]) spaces with m >= N(0) and alpha >= omega(1) are the trivial ones. This result leads to some elementary questions on large cardinals.

The Interval Constructor on classes of ML-algebras

Monoidal logic, ML for short, which formalized the fuzzy logics of continuous t-norms and their residua, has arisen great interest, since it has been applied to fuzzy mathematics, artificial intelligence, and other areas. It is clear that fuzzy logics basically try to represent imperfect or fuzzy information aiming to model the natural human reasoning. On the other hand, in order to deal with imprecision in the computational representation of real numbers, the use of intervals have been proposed, as it can guarantee that the results of numerical computation are in a bounded interval, controlling, in this way, the numerical errors produced by successive roundings. There are several ways to connect both areas; the most usual one is to consider interval membership degrees. The algebraic counterpart of ML is ML-algebra, an interesting structure due to the fact that by adding some properties it is possible to reach different classes of residuated lattices. We propose to apply an interval constructor to ML-algebras and some of their subclasses, to verify some properties within these algebras, in addition to the analysis of the algebraic aspects of them

Formação de classes de equivalência através de reforçamento específico em crianças com dificuldades de aprendizagem

Na Análise do Comportamento, vários estudos são realizados a fim de entender como comportamentos podem produtivamente ser controlados por eventos arbitrariamente relacionados, através da formação classes de equivalência. A inclusão de estímulos reforçadores nas classes tem sido apontada como um possível facilitador de sua formação. O presente estudo teve como objetivos avaliar a formação de classes de equivalência mediada por estímulos reforçadores específicos em crianças que apresentam baixo rendimento escolar. Usando crianças com desenvolvimento típico e em maior número, comparado com a literatura, pretendeu-se obter dados com menor variabilidade intersujeitos que é comumente encontrada nesse tipo de pesquisa. Para isso, foram utilizados reforçadores específicos com quatorze crianças (no Experimento I) que apresentam dificuldades de aprendizagem. O procedimento do Experimento 1 foi dividido em 10 fases. Em todas as fases, houve reforçadores específicos (frutas ou brinquedos) para cada uma das classes potenciais que se pretendia verificar. Inicialmente foi realizado um treino de pareamento por identidade com os estímulos dos Conjuntos A (A I e A2), B (B 1 e B2) e C (C I e C2) com reforçamento contínuo, seguido do mesmo treino com Reforçamento Intermitente. Logo após esses treinos foram realizados os testes de relações emergentes AB/BA, ACICA e BCICB. Antes de cada teste foi feito o retorno às discriminações de linha de base. Os dados do Experimento I evidenciam grande variabilidade intersujeitos nos testes de formação de classes. O Experimento 2 pretendeu investigar o efeito de dois tipos de pré-treino sobre o desempenho nos testes de formação de classes. Foi realizado com seis crianças e subdividido em dois grupos. O Grupo I foi submetido a um pré-treino de pareamento por identidade e o Grupo 2 a um pré-treino de pareamento arbitrário. Os resultados confirmam parcialmente a hipótese de que pré-treino de pareamento arbitrário pode reduzir a variabilidade inter-sujeitos nesse tipo estudos, pois altas taxas de variabilidade persistem no presente estudo. .Estudos posteriores deverão explorar essa possibilidade mais sistematicamente.

Diastolic function and functional capacity after a single session of continuous positive airway pressure in patients with compensated heart failure

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Limit Cycles for a Class of Continuous and Discontinuous Cubic Polynomial Differential Systems

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

On the Limit Cycles for a Class of Continuous Piecewise Linear Differential Systems with Three Zones

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Opportunistic pathogens and elements of the resistome that are common in bottled mineral water support the need for continuous surveillance

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Pseudodifferential operators on Hilbert space riggings with associated psi *-algebras and generalized Hörmander classes

The present thesis is concerned with certain aspects of differential and pseudodifferential operators on infinite dimensional spaces. We aim to generalize classical operator theoretical concepts of pseudodifferential operators on finite dimensional spaces to the infinite dimensional case. At first we summarize some facts about the canonical Gaussian measures on infinite dimensional Hilbert space riggings. Considering the naturally unitary group actions in $L^2(H_-,gamma)$ given by weighted shifts and multiplication with $e^{iSkp{t}{cdot}_0}$ we obtain an unitary equivalence $F$ between them. In this sense $F$ can be considered as an abstract Fourier transform. We show that $F$ coincides with the Fourier-Wiener transform. Using the Fourier-Wiener transform we define pseudodifferential operators in Weyl- and Kohn-Nirenberg form on our Hilbert space rigging. In the case of this Gaussian measure $gamma$ we discuss several possible Laplacians, at first the Ornstein-Uhlenbeck operator and then pseudo-differential operators with negative definite symbol. In the second case, these operators are generators of $L^2_gamma$-sub-Markovian semi-groups and $L^2_gamma$-Dirichlet-forms. In 1992 Gramsch, Ueberberg and Wagner described a construction of generalized Hörmander classes by commutator methods. Following this concept and the classical finite dimensional description of $Psi_{ro,delta}^0$ ($0leqdeltaleqroleq 1$, $delta< 1$) in the $C^*$-algebra $L(L^2)$ by Beals and Cordes we construct in both cases generalized Hörmander classes, which are $Psi^*$-algebras. These classes act on a scale of Sobolev spaces, generated by our Laplacian. In the case of the Ornstein-Uhlenbeck operator, we prove that a large class of continuous pseudodifferential operators considered by Albeverio and Dalecky in 1998 is contained in our generalized Hörmander class. Furthermore, in the case of a Laplacian with negative definite symbol, we develop a symbolic calculus for our operators. We show some Fredholm-criteria for them and prove that these Fredholm-operators are hypoelliptic. Moreover, in the finite dimensional case, using the Gaussian-measure instead of the Lebesgue-measure the index of these Fredholm operators is still given by Fedosov's formula. Considering an infinite dimensional Heisenberg group rigging we discuss the connection of some representations of the Heisenberg group to pseudo-differential operators on infinite dimensional spaces. We use this connections to calculate the spectrum of pseudodifferential operators and to construct generalized Hörmander classes given by smooth elements which are spectrally invariant in $L^2(H_-,gamma)$. Finally, given a topological space $X$ with Borel measure $mu$, a locally compact group $G$ and a representation $B$ of $G$ in the group of all homeomorphisms of $X$, we construct a Borel measure $mu_s$ on $X$ which is invariant under $B(G)$.

Mixture modeling for joint analysis of survival, discrete, and continuous data

Mixture modeling is commonly used to model categorical latent variables that represent subpopulations in which population membership is unknown but can be inferred from the data. In relatively recent years, the potential of finite mixture models has been applied in time-to-event data. However, the commonly used survival mixture model assumes that the effects of the covariates involved in failure times differ across latent classes, but the covariate distribution is homogeneous. The aim of this dissertation is to develop a method to examine time-to-event data in the presence of unobserved heterogeneity under a framework of mixture modeling. A joint model is developed to incorporate the latent survival trajectory along with the observed information for the joint analysis of a time-to-event variable, its discrete and continuous covariates, and a latent class variable. It is assumed that the effects of covariates on survival times and the distribution of covariates vary across different latent classes. The unobservable survival trajectories are identified through estimating the probability that a subject belongs to a particular class based on observed information. We applied this method to a Hodgkin lymphoma study with long-term follow-up and observed four distinct latent classes in terms of long-term survival and distributions of prognostic factors. Our results from simulation studies and from the Hodgkin lymphoma study demonstrated the superiority of our joint model compared with the conventional survival model. This flexible inference method provides more accurate estimation and accommodates unobservable heterogeneity among individuals while taking involved interactions between covariates into consideration.^