Numerical Analysis and Scientific Computing Commons^™

Large Scale Online Kernel Classification, Jialei Wang, Peilin Zhao, Steven C. H. Hoi, Jinfeng Zhuang, Zhi-Yong Liu

Research Collection School Of Computing and Information Systems

In this work, we present a new framework for large scale online kernel classification, making kernel methods efficient and scalable for large-scale online learning tasks. Unlike the regular budget kernel online learning scheme that usually uses different strategies to bound the number of support vectors, our framework explores a functional approximation approach to approximating a kernel function/matrix in order to make the subsequent online learning task efficient and scalable. Specifically, we present two different online kernel machine learning algorithms: (i) the Fourier Online Gradient Descent (FOGD) algorithm that applies the random Fourier features for approximating kernel functions; and (ii) the …

Mkboost: A Framework Of Multiple Kernel Boosting, Hao Xia, Steven C. H. Hoi

Research Collection School Of Computing and Information Systems

Multiple kernel learning (MKL) is a promising family of machine learning algorithms using multiple kernel functions for various challenging data mining tasks. Conventional MKL methods often formulate the problem as an optimization task of learning the optimal combinations of both kernels and classifiers, which usually results in some forms of challenging optimization tasks that are often difficult to be solved. Different from the existing MKL methods, in this paper, we investigate a boosting framework of MKL for classification tasks, i.e., we adopt boosting to solve a variant of MKL problem, which avoids solving the complicated optimization tasks. Specifically, we present …

Contributions To The Cuore Collaboration, Samuel Joseph Meijer

Physics

This paper describes work done between 2010 and 2013 to contribute to the CUORE collaboration, a physics collaboration searching for neutrinoless double-beta decay in tellurium. Measurement of this decay would indicate fundamental information about the nature of the neutrino. The implementation of a parylene-coated detector frame is described. Also, a temperature stabilization system for an automated gluing system was constructed. An image recognition algorithm is described for locating spots of glue and evaluating their acceptability.

Reviving Dormant Ties In An Online Social Network Experiment, Ee Peng Lim, Denzil Correa, David Lo, Michael Finegold, Feida Zhu

Research Collection School Of Computing and Information Systems

Social network users connect and interact with one another to fulfil different kinds of social and information needs. When interaction ceases between two users, we say that their tie becomes dormant. While there are different underlying reasons of dormant ties, it is important to find means to revive such ties so as to maintain vibrancy in the relationships. In this work, we thus focus on designing an online experiment to evaluate the effectiveness of personalized social messages to revive dormant ties. The experiment carefully selects users with dormant ties so that no user gets mixed treatments and be affected by …

Shortlisting Top-K Assignments, Yimin Lin, Kyriakos Mouratidis

Research Collection School Of Computing and Information Systems

In this paper we identify a novel query type, the top-K assignment query (αTop-K). Consider a set of objects and a set of suppliers, where each object must be assigned to one supplier. Assume that there is a cost associated with every object-supplier pair. If we allocate each object to the server with the smallest cost (for the specific object), the derived overall assignment will have the minimum total cost. In many scenarios, however, runner-up assignments may be required too, like for example when a decision maker needs to make additional considerations, not captured by individual object-supplier costs. In this …

Active Learning With Expert Advice, Peilin Zhao, Steven C. H. Hoi, Jinfeng Zhuang

Research Collection School Of Computing and Information Systems

Conventional learning with expert advice methods assumes a learner is always receiving the outcome (e.g., class labels) of every incoming training instance at the end of each trial. In real applications, acquiring the outcome from oracle can be costly or time consuming. In this paper, we address a new problem of active learning with expert advice, where the outcome of an instance is disclosed only when it is requested by the online learner. Our goal is to learn an accurate prediction model by asking the oracle the number of questions as small as possible. To address this challenge, we propose …

Mining Direct Antagonistic Communities In Signed Social Networks, David Lo, Didi Surian, Philips Kokoh Prasetyo, Zhang Kuan, Ee Peng Lim

Research Collection School Of Computing and Information Systems

Social networks provide a wealth of data to study relationship dynamics among people. Most social networks such as Epinions and Facebook allow users to declare trusts or friendships with other users. Some of them also allow users to declare distrusts or negative relationships. When both positive and negative links co-exist in a network, some interesting community structures can be studied. In this work, we mine Direct Antagonistic Communities (DACs) within such signed networks. Each DAC consists of two sub-communities with positive relationships among members of each sub-community, and negative relationships among members of the other sub-community. Identifying direct antagonistic communities …

Using Correlated Subset Structure For Compressive Sensing Recovery, Atul Divekar, Deanna Needell

CMC Faculty Publications and Research

Compressive sensing is a methodology for the reconstruction of sparse or compressible signals using far fewer samples than required by the Nyquist criterion. However, many of the results in compressive sensing concern random sampling matrices such as Gaussian and Bernoulli matrices. In common physically feasible signal acquisition and reconstruction scenarios such as super-resolution of images, the sensing matrix has a non-random structure with highly correlated columns. Here we present a compressive sensing recovery algorithm that exploits this correlation structure. We provide algorithmic justification as well as empirical comparisons.

Brovine: Mammary Gland Gene Database, Therin C. Irwin

Computer Science and Software Engineering

Brovine is used by the Animal Science department at Cal Poly to catalog and analyze genetic information. Brovine, or the Mammary Gland Gene Database, is a system used to store and categorize genetic information which is gathered through experimentation and through TESS, a web application that lets users search through catalogs of similar genetic information. This document describes the purpose, use, and maintenance of Brovine.

Pin: Measuring Asymmetric Information In Financial Markets With R, Paolo Zagaglia

The R Journal

The package PIN computes a measure of asymmetric information in financial markets, the so-called probability of informed trading. This is obtained from a sequential trade model and is used to study the determinants of an asset price. Since the probability of informed trading depends on the number of buy- and sell-initiated trades during a trading day, this paper discusses the entire modelling cycle, from data handling to the computation of the probability of informed trading and the estimation of parameters for the underlying theoretical model.

Statistical Software From A Blind Person's Perspective, A. Jonathan R. Godfrey

The R Journal

Blind people have experienced access issues to many software applications since the advent of the Windows operating system; statistical software has proven to follow the rule and not be an exception. The ability to use R within minutes of download with next to no adaptation has opened doors for accessible production of statistical analyses for this author (himself blind) and blind students around the world. This article shows how little is required to make R the most accessible statistical software available today. There is any number of ramifications that this opportunity creates for blind students, especially in terms of their …

Multiple Factor Analysis For Contingency Tables In The Factominer Package, Belchin Kostov, Mónica Bécue-Bertaut, François Husson

The R Journal

We present multiple factor analysis for contingency tables (MFACT) and its implementation in the FactoMineR package. This method, through an option of the MFA function, allows us to deal with multiple contingency or frequency tables, in addition to the categorical and quantitative multiple tables already considered in previous versions of the package. Thanks to this revised function, either a multiple contingency table or a mixed multiple table integrating quantitative, categorical and frequency data can be tackled.

The FactoMineR package (Lê et al., 2008; Husson et al., 2011) offers the most commonly used principal component methods: principal component analysis (PCA), correspondence …

Generalized Simulated Annealing For Global Optimization: The Gensa Package, Yang Xiang, Sylvain Gubian, Brain Suomela, Julia Hoeng

The R Journal

Many problems in statistics, finance, biology, pharmacology, physics, mathematics, economics, and chemistry involve determination of the global minimum of multidimensional functions. R packages for different stochastic methods such as genetic algorithms and differential evolution have been developed and successfully used in the R community. Based on Tsallis statistics, the R package GenSA was developed for generalized simulated annealing to process complicated non-linear objective functions with a large number of local minima. In this paper we provide a brief introduction to the R package and demonstrate its utility by solving a non-convex portfolio optimization problem in finance and the Thomson problem …

R Foundation News, Kurt Hornik

The R Journal

New Benefactors

Quartz, Bio, Switzerland

New supporting Institutions

Institute for Geoinformatics, Westfälische Wilhelms-Universität Münster, Germany

News From The Bioconductor Project, Bioconductor Team

The R Journal

Bioconductor 2.12 was released on 3 October 2012. It is compatible with R 3.0.1, and consists of 671 software packages and more than 675 up-to-date annotation packages. The release includes 65 new software packages, and enhancements to many others. Descriptions of new packages and updated NEWS files provided by current package maintainers are at http://bioconductor.org/news/bioc_2_12_release/.

Conference Review: The 6th Chinese R Conference, Jing Leng, Jingjing Guan

The R Journal

The 6th Chinese R Conference (Beijing session) was held in the Sinology Pavilion of Renmin University of China (RUC), Beijing, from May 18th to 19th, 2013. The conference was orga nized by the “Capital of Statistics” (COS, http://cos.name), an online statistical community in China. It was sponsored and co-organized by the Center for Applied Statistics of RUC, the School of Statistics of RUC, and the Business Intelligence Research Center of Peking University

Possible Directions For Improving Dependency Versioning In R, Jeroen Ooms

The R Journal

One of the most powerful features of R is its infrastructure for contributed code. The built-in package manager and complementary repositories provide a great system for development and exchange of code, and have played an important role in the growth of the platform towards the de-facto standard in statistical computing that it is today. However, the number of packages on CRAN and other repositories has increased beyond what might have been foreseen, and is revealing some limitations of the current design. One such problem is the general lack of dependency versioning in the infrastructure. This paper explores this problem in …

Beadarrayfilter: An R Package To Filter Beads, Anyiawung Chiara Forcheh, Geert Verbeke, Adetayo Kasim, Dan Lin, Ziv Shkedy, Willem Talloen, Hinrich W.H. Göhlmann, Lieven Clement

The R Journal

Microarrays enable the expression levels of thousands of genes to be measured simultaneously. However, only a small fraction of these genes are expected to be expressed under different experimental conditions. Nowadays, filtering has been introduced as a step in the microarray pre-processing pipeline. Gene filtering aims at reducing the dimensionality of data by filtering redundant features prior to the actual statistical analysis. Previous filtering methods focus on the Affymetrix platform and can not be easily ported to the Illumina platform. As such, we developed a filtering method for Illumina bead arrays. We developed an R package, beadarrayFilter, to implement …

Ggmap: Spatial Visualization With Ggplot2, David Kahle, Hadley Wickham

The R Journal

In spatial statistics the ability to visualize data and models super imposed with their basic social landmarks and geographic context is in valuable. ggmap is a new tool which enables such visualization by combining the spatial information of static maps from Google Maps,Open Street Map, Stamen Maps or Cloud Made Maps with the layered grammar of graphics implementation of ggplot2. In addition, several new utility functions are introduced which allow the user to access the Google Geocoding, Distance Matrix, and Directions APIs. The result is an easy, consistent and modular frame work for spatial graphics with several convenient tools …

Let Graphics Tell The Story: Datasets In R, Antony Unwin, Heike Hofmann, Dianne Cook

The R Journal

Graphics are good for showing the information in datasets and for complementing modelling. Sometimes graphics show information models miss, sometimes graphics help to make model results more understandable, and sometimes models show whether information from graphics has statistical support or not. It is the interplay of the two approaches that is valuable. Graphics could be used a lot more in R examples and we explore this idea with some datasets available in R packages.