Connecting social media to e-commerce:cold-start product recommendation using microblogging information

In recent years, the boundaries between e-commerce and social networking have become increasingly blurred. Many e-commerce websites support the mechanism of social login where users can sign on the websites using their social network identities such as their Facebook or Twitter accounts. Users can also post their newly purchased products on microblogs with links to the e-commerce product web pages. In this paper, we propose a novel solution for cross-site cold-start product recommendation, which aims to recommend products from e-commerce websites to users at social networking sites in 'cold-start' situations, a problem which has rarely been explored before. A major challenge is how to leverage knowledge extracted from social networking sites for cross-site cold-start product recommendation. We propose to use the linked users across social networking sites and e-commerce websites (users who have social networking accounts and have made purchases on e-commerce websites) as a bridge to map users' social networking features to another feature representation for product recommendation. In specific, we propose learning both users' and products' feature representations (called user embeddings and product embeddings, respectively) from data collected from e-commerce websites using recurrent neural networks and then apply a modified gradient boosting trees method to transform users' social networking features into user embeddings. We then develop a feature-based matrix factorization approach which can leverage the learnt user embeddings for cold-start product recommendation. Experimental results on a large dataset constructed from the largest Chinese microblogging service Sina Weibo and the largest Chinese B2C e-commerce website JingDong have shown the effectiveness of our proposed framework.

Evaluating statistical and machine learning methods to predict risk of in-hospital child mortality in Uganda

Thesis (Master's)--University of Washington, 2016-08

Essays on Machine Learning and Hedonic Models

Thesis (Ph.D.)--University of Washington, 2016-08

Sparse hierarchical Bayesian models for detecting relevant antigenic sites in virus evolution

Understanding how virus strains offer protection against closely related emerging strains is vital for creating effective vaccines. For many viruses, including Foot-and-Mouth Disease Virus (FMDV) and the Influenza virus where multiple serotypes often co-circulate, in vitro testing of large numbers of vaccines can be infeasible. Therefore the development of an in silico predictor of cross-protection between strains is important to help optimise vaccine choice. Vaccines will offer cross-protection against closely related strains, but not against those that are antigenically distinct. To be able to predict cross-protection we must understand the antigenic variability within a virus serotype, distinct lineages of a virus, and identify the antigenic residues and evolutionary changes that cause the variability. In this thesis we present a family of sparse hierarchical Bayesian models for detecting relevant antigenic sites in virus evolution (SABRE), as well as an extended version of the method, the extended SABRE (eSABRE) method, which better takes into account the data collection process. The SABRE methods are a family of sparse Bayesian hierarchical models that use spike and slab priors to identify sites in the viral protein which are important for the neutralisation of the virus. In this thesis we demonstrate how the SABRE methods can be used to identify antigenic residues within different serotypes and show how the SABRE method outperforms established methods, mixed-effects models based on forward variable selection or l1 regularisation, on both synthetic and viral datasets. In addition we also test a number of different versions of the SABRE method, compare conjugate and semi-conjugate prior specifications and an alternative to the spike and slab prior; the binary mask model. We also propose novel proposal mechanisms for the Markov chain Monte Carlo (MCMC) simulations, which improve mixing and convergence over that of the established component-wise Gibbs sampler. The SABRE method is then applied to datasets from FMDV and the Influenza virus in order to identify a number of known antigenic residue and to provide hypotheses of other potentially antigenic residues. We also demonstrate how the SABRE methods can be used to create accurate predictions of the important evolutionary changes of the FMDV serotypes. In this thesis we provide an extended version of the SABRE method, the eSABRE method, based on a latent variable model. The eSABRE method takes further into account the structure of the datasets for FMDV and the Influenza virus through the latent variable model and gives an improvement in the modelling of the error. We show how the eSABRE method outperforms the SABRE methods in simulation studies and propose a new information criterion for selecting the random effects factors that should be included in the eSABRE method; block integrated Widely Applicable Information Criterion (biWAIC). We demonstrate how biWAIC performs equally to two other methods for selecting the random effects factors and combine it with the eSABRE method to apply it to two large Influenza datasets. Inference in these large datasets is computationally infeasible with the SABRE methods, but as a result of the improved structure of the likelihood, we are able to show how the eSABRE method offers a computational improvement, leading it to be used on these datasets. The results of the eSABRE method show that we can use the method in a fully automatic manner to identify a large number of antigenic residues on a variety of the antigenic sites of two Influenza serotypes, as well as making predictions of a number of nearby sites that may also be antigenic and are worthy of further experiment investigation.

Um estudo de descritores e classificadores para reconhecimento de bagas de café.

RESUMO - Métodos de reconhecimento de frutos baseados na utilização de diferentes descritores e classificadores foram estudados. Foi utilizada uma base de dados de 3.393 imagens de café e não-café anteriormente criada e rotulada manualmente. Testes quantitativos demonstraram a identificação de bagas com 93% de precisão e 77% de cobertura utilizando descritores HoG adicionados a mediana dos componentes de cor do formato La*b*, aliados ao classificador Gradient Boosting. Esses resultados melhoram o método anteriormente proposto por Santos (2015), e demonstram a possibilidade de evolução de métodos que podem ser aplicados em metodologias de agricultura de precisão, monitoramento e predição de safra.

Boosting the effects of a curriculum based injury prevention program through a school connectedness intervention : final report to NRMA-ACT Road Safety Trust

Injury is the leading cause of death among young people (AIHW, 2008). A primary contributing factor to injury among adolescents is risk taking behaviour, including road related risks such as risky bicycle and motorcycle use and riding with dangerous or drink-drivers. Injury rates increase dramatically throughout adolescence, at the same time as adolescents are becoming more involved in risk taking behaviour. Also throughout this period, adolescents‟ connectedness to school is decreasing (Monahan, Oesterle & Hawkins, 2010; Whitlock, 2004). School connectedness refers to „the extent to which students feel personally accepted, respected, included, and supported by others in the school‟ (Goodenow, 1993, p. 80), and has been repeatedly shown to be a critical protective factor in adolescent development. For example, school connectedness has been shown to be associated with decreased risk taking behaviour, including violence and alcohol and other drug use (e.g., Resnick et al., 1997), as well as with decreased transport risk taking and vehicle related injuries (Chapman et al., accepted April 2011). This project involved the pilot evaluation of a school connectedness intervention (a professional development program for teachers) to reduce adolescent risk taking behaviour and injury. This intervention has been developed for use as a component of the Skills for Preventing Injury in Youth (SPIY) curriculum based injury prevention program for early adolescents. The objectives of this research were to: 1. Implement a trial School Connectedness intervention (professional development program for teachers) in ACT high schools, and evaluate using comparison high schools. 2. Determine whether the School Connectedness program impacts on adolescent risk taking behaviour and associated injuries (particularly transport risks and injuries). 3. Evaluate the process effectiveness of the School Connectedness program.

I. The structure and properties of closed circular duplex DNA. II. Chemically interacting systems at equilibrium in a buoyant density gradient

I. The binding of the intercalating dye ethidium bromide to closed circular SV 40 DNA causes an unwinding of the duplex structure and a simultaneous and quantitatively equivalent unwinding of the superhelices. The buoyant densities and sedimentation velocities of both intact (I) and singly nicked (II) SV 40 DNAs were measured as a function of free dye concentration. The buoyant density data were used to determine the binding isotherms over a dye concentration range extending from 0 to 600 µg/m1 in 5.8 M CsCl. At high dye concentrations all of the binding sites in II, but not in I, are saturated. At free dye concentrations less than 5.4 µg/ml, I has a greater affinity for dye than II. At a critical amount of dye bound I and II have equal affinities, and at higher dye concentration I has a lower affinity than II. The number of superhelical turns, τ, present in I is calculated at each dye concentration using Fuller and Waring's (1964) estimate of the angle of duplex unwinding per intercalation. The results reveal that SV 40 DNA I contains about -13 superhelical turns in concentrated salt solutions.

The free energy of superhelix formation is calculated as a function of τ from a consideration of the effect of the superhelical turns upon the binding isotherm of ethidium bromide to SV 40 DNA I. The value of the free energy is about 100 kcal/mole DNA in the native molecule. The free energy estimates are used to calculate the pitch and radius of the superhelix as a function of the number of superhelical turns. The pitch and radius of the native I superhelix are 430 Å and 135 Å, respectively.

A buoyant density method for the isolation and detection of closed circular DNA is described. The method is based upon the reduced binding of the intercalating dye, ethidium bromide, by closed circular DNA. In an application of this method it is found that HeLa cells contain in addition to closed circular mitochondrial DNA of mean length 4.81 microns, a heterogeneous group of smaller DNA molecules which vary in size from 0.2 to 3.5 microns and a paucidisperse group of multiples of the mitochondrial length.

II. The general theory is presented for the sedimentation equilibrium of a macromolecule in a concentrated binary solvent in the presence of an additional reacting small molecule. Equations are derived for the calculation of the buoyant density of the complex and for the determination of the binding isotherm of the reagent to the macrospecies. The standard buoyant density, a thermodynamic function, is defined and the density gradients which characterize the four component system are derived. The theory is applied to the specific cases of the binding of ethidium bromide to SV 40 DNA and of the binding of mercury and silver to DNA.

Generalized power method for sparse principal component analysis

In this paper we develop a new approach to sparse principal component analysis (sparse PCA). We propose two single-unit and two block optimization formulations of the sparse PCA problem, aimed at extracting a single sparse dominant principal component of a data matrix, or more components at once, respectively. While the initial formulations involve nonconvex functions, and are therefore computationally intractable, we rewrite them into the form of an optimization program involving maximization of a convex function on a compact set. The dimension of the search space is decreased enormously if the data matrix has many more columns (variables) than rows. We then propose and analyze a simple gradient method suited for the task. It appears that our algorithm has best convergence properties in the case when either the objective function or the feasible set are strongly convex, which is the case with our single-unit formulations and can be enforced in the block case. Finally, we demonstrate numerically on a set of random and gene expression test problems that our approach outperforms existing algorithms both in quality of the obtained solution and in computational speed. © 2010 Michel Journée, Yurii Nesterov, Peter Richtárik and Rodolphe Sepulchre.

Optimization on the orthogonal group for independent component analysis

This paper derives a new algorithm that performs independent component analysis (ICA) by optimizing the contrast function of the RADICAL algorithm. The core idea of the proposed optimization method is to combine the global search of a good initial condition with a gradient-descent algorithm. This new ICA algorithm performs faster than the RADICAL algorithm (based on Jacobi rotations) while still preserving, and even enhancing, the strong robustness properties that result from its contrast. © Springer-Verlag Berlin Heidelberg 2007.

Soil microbial community analysis using denaturing gradient gel electrophoresis

The most biological diversity on this planet is probably harbored in soils. Understanding the diversity and function of the microbiological component of soil poses great challenges that are being overcome by the application of molecular biological approaches. This review covers one of many approaches being used: separation of polymerase chain reaction (PCR) amplicons using denaturing gradient gel electrophoresis (DGGE). Extraction of nucleic acids directly from soils allows the examination of a community without the limitation posed by cultivation. Polymerase chain reaction provides a means to increase the numbers of a target for its detection on gels. Using the rRNA genes as a target for PCR provides phylogenetic information on populations comprising communities. Fingerprints produced by this method have allowed spatial and temporal comparisons of soil communities within and between locations or among treatments. Numerous samples can be compared because of the rapid high throughput nature of this method. Scientists now have the means to begin addressing complex ecological questions about the spatial, temporal, and nutritional interactions faced by microbes in the soil environment.

Total analytical method for Radix astragali extract using two-binary multi-segment gradient elution liquid chromatography

There is an urgent need for thorough analysis of Radix astragali, a widely used Chinese herb, for quality control purposes. This paper describes the development of a total analytical method for Radix astragali extract, a multi-component complex mixture. Twenty-four components were separated step by step from the extract using a series of isocratic isopropanol-methanol elutions, and then 42 components were separated similarly using methanol-water elutions. Based on the log k(w) and -S of the 66 components obtained from the above procedure and the optimization software developed in our laboratory, an optimum elution program consisting of seven methanol-water segments and four isopropanol-methanol segments was developed to finish the task of analyzing the total components in a single run. Under optimized gradient conditions, the sample of Radix astragali extract was analyzed. As expected, most of the components were well separated and the experimental chromatogram was in a good agreement with the predicted one.

Improved description of chemical barriers with generalized gradient approximations (GGAs) and meta-GGAs

The performance of exchange and correlation (xc) functionals of the generalized gradient approximation (GGA) type and of the meta-GGA type in the calculation of chemical reactions is related to topological features of the electron density which, in turn, are connected to the orbital structure of chemical bonds within the Kohn-Sham (KS) theory. Seventeen GGA and meta-GGA xc functionals are assessed for 15 hydrogen abstraction reactions and 3 symmetrical S(N)2 reactions. Systems that are problematic for standard GGAs characteristically have enhanced values of the dimensionless gradient argument s(sigma)(2) with local maxima in the bonding region. The origin of this topological feature is the occupation of valence KS orbitals with an antibonding or essentially nonbonding character. The local enhancement of s(sigma)(2) yields too negative exchange-correlation energies with standard GGAs for the transition state of the S(N)2 reaction, which leads to the reduced calculated reaction barriers. The unwarranted localization of the effective xc hole of the standard GGAs, i.e., the nondynamical correlation that is built into them but is spurious in this case, wields its effect by their s(sigma)(2) dependence. Barriers are improved for xc functionals with the exchange functional OPTX as x component, which has a modified dependence on s(sigma)(2). Standard GGAs also underestimate the barriers for the hydrogen abstraction reactions. In this case the barriers are improved by correlation functionals, such as the Laplacian-dependent (LAP3) functional, which has a modified dependence on the Coulomb correlation of the opposite- and like-spin electrons. The best overall performance is established for the combination OLAP3 of OPTX and LAP3.

Projected gradient approach to the numerical solution of the SCoTLASS

The SCoTLASS problem-principal component analysis modified so that the components satisfy the Least Absolute Shrinkage and Selection Operator (LASSO) constraint-is reformulated as a dynamical system on the unit sphere. The LASSO inequality constraint is tackled by exterior penalty function. A globally convergent algorithm is developed based on the projected gradient approach. The algorithm is illustrated numerically and discussed on a well-known data set. (c) 2004 Elsevier B.V. All rights reserved.

Effects of oxidised low density lipoprotein on dendritic cells: a possible immunoregulatory component of the atherogenic micro-environment?

Objective: The objective of this study was to explore the relationship between low density lipoprotein (LDL) and dendritic cell (DC) activation, based upon the hypothesis that reactive oxygen species (ROS)-mediated modification of proteins that may be present in local DC microenvironments could be important as mediators of this activation. Although LDL are known to be oxidised in vivo, and taken up by macrophages during atherogenesis; their effect on DC has not been explored previously. Methods: Human DCs were prepared from peripheral blood monocytes using GM-CSF and IL-4. Plasma LDLs were isolated by sequential gradient centrifugation, oxidised in CuSO4, and oxidation arrested to yield mild, moderate and highly oxidised LDL forms. DCs exposed to these LDLs were investigated using combined phenotypic, functional (autologous T cell activation), morphological and viability assays. Results: Highly-oxidised LDL increased DC HLA-DR, CD40 and CD86 expression, corroborated by increased DC-induced T cell proliferation. Both native and oxidised LDL induced prominent DC clustering. However, high concentrations of highly-oxidised LDL inhibited DC function, due to increased DC apoptosis. Conclusions: This study supports the hypothesis that oxidised LDL are capable of triggering the transition from sentinel to messenger DC. Furthermore, the DC clustering–activation–apoptosis sequence in the presence of different LDL forms is consistent with a regulatory DC role in immunopathogenesis of atheroma. A sequence of initial accumulation of DC, increasing LDL oxidation, and DC-induced T cell activation, may explain why local breach of tolerance can occur. Above a threshold level, however, supervening DC apoptosis limits this, contributing instead to the central plaque core.