Tree-based Visualization Research Articles

A Visual Mining Approach to Improved Multiple- Instance Learning

Multiple-instance learning (MIL) is a paradigm of machine learning that aims to classify a set (bag) of objects (instances), assigning labels only to the bags. This problem is often addressed by selecting an instance to represent each bag, transforming an MIL problem into standard supervised learning. Visualization can be a useful tool to assess learning scenarios by incorporating the users’ knowledge into the classification process. Considering that multiple-instance learning is a paradigm that cannot be handled by current visualization techniques, we propose a multiscale tree-based visualization called MILTree to support MIL problems. The first level of the tree represents the bags, and the second level represents the instances belonging to each bag, allowing users to understand the MIL datasets in an intuitive way. In addition, we propose two new instance selection methods for MIL, which help users improve the model even further. Our methods can handle both binary and multiclass scenarios. In our experiments, SVM was used to build the classifiers. With support of the MILTree layout, the initial classification model was updated by changing the training set, which is composed of the prototype instances. Experimental results validate the effectiveness of our approach, showing that visual mining by MILTree can support exploring and improving models in MIL scenarios and that our instance selection methods outperform the currently available alternatives in most cases.

Reduced bonobo MHC class I diversity predicts a reduced viral peptide binding ability compared to chimpanzees

BackgroundThe highly polymorphic genes of the major histocompatibility complex (MHC) class I are involved in defense against viruses and other intracellular pathogens. Although several studies found reduced MHC class I diversity in bonobos in comparison to the closely related chimpanzee, it is unclear if this lower diversity also influences the functional ability of MHC class I molecules in bonobos. Here, we use a bioinformatic approach to analyze the viral peptide binding ability of all published bonobo MHC class I molecules (n = 58) in comparison to all published chimpanzee MHC class I molecules (n = 161) for the class I loci A, B, C and A-like.ResultsWe examined the peptide binding ability of all 219 different MHC class I molecules to 5,788,712 peptides derived from 1432 different primate viruses and analyzed the percentage of bound peptides and the overlap of the peptide binding repertoires of the two species. We conducted multiple levels of analysis on the “species”-, “population”- and “individual”-level to account for the characterization of MHC variation in a larger number of chimpanzees and their broader geographic distribution. We found a lower percentage of bound peptides in bonobos at the B locus in the “population”-level comparison and at the B and C loci in the “individual”-level comparison. Furthermore, we found evidence of a limited peptide binding repertoire in bonobos by tree-based visualization of functional clustering of MHC molecules, as well as an analysis of peptides bound by both species.ConclusionOur results suggest a reduced MHC class I viral peptide binding ability at the B and C loci in bonobos compared to chimpanzees. The effects of this finding on the immune defense against viruses in wild living bonobos are unclear. However, special caution is needed to prevent introduction and spread of new viruses to bonobos, as their defensive ability to cope with new viruses could be limited compared to chimpanzees.

A logic-based representation and tree-based visualization method for building regulatory requirements

BackgroundMany research and development efforts have been made for automated compliance checking of building designs with regulatory requirements, but there is a lack of a non-proprietary and user-understandable representation of building regulations to support automated compliance checking in the construction domain that is experimentally tested for understandability and reading speed.MethodsThis research investigates a logic-based representation and tree-based visualization method for building regulatory requirements. The logic-based representation is based on classic logic programming language and can directly support automated compliance checking. The tree-based visualization is expected to improve the understandability and reading speed of the logic-based representation. Therefore, this method attempts to add to the limited research in non-proprietary and user-understandable representation of building regulations that is experimentally tested for understandability and reading speed. To test the understandability and reading speed of regulatory requirements using this representation and visualization method, a survey was conducted to compare different representations, namely, text, logic-based, and tree-based.ResultsStatistical analysis of the survey results shows that the proposed tree-based visualization method can significantly improve the understandability and reading speed of the logic-based regulatory requirement representation and this visualization method is at a comparable status with the original text representation of regulatory requirements in terms of understandability and reading speed.Conclusions(1) The investigated logic-based representation and tree-based visualization method for regulatory requirements serves as one potential non-proprietary and user-understandable representation of building regulations; (2) this research shows that computable representations of regulatory requirements can achieve understandability and reading speed that are comparable to the original text representation through tree-based visualization; and (3) this research reveals that the tree-based visualization of regulatory requirements improves the understandability and reading speed of regulatory requirements when such use is compared to the computable logic-based representation.

Tree-Based Visualization and Optimization for Image Collection.

The visualization of an image collection is the process of displaying a collection of images on a screen under some specific layout requirements. This paper focuses on an important problem that is not well addressed by the previous methods: visualizing image collections into arbitrary layout shapes while arranging images according to user-defined semantic or visual correlations (e.g., color or object category). To this end, we first propose a property-based tree construction scheme to organize images of a collection into a tree structure according to user-defined properties. In this way, images can be adaptively placed with the desired semantic or visual correlations in the final visualization layout. Then, we design a two-step visualization optimization scheme to further optimize image layouts. As a result, multiple layout effects including layout shape and image overlap ratio can be effectively controlled to guarantee a satisfactory visualization. Finally, we also propose a tree-transfer scheme such that visualization layouts can be adaptively changed when users select different "images of interest." We demonstrate the effectiveness of our proposed approach through the comparisons with state-of-the-art visualization techniques.

MaxdLoad2 and maxdBrowse: standards-compliant tools for microarray experimental annotation, data management and dissemination

BackgroundmaxdLoad2 is a relational database schema and Java® application for microarray experimental annotation and storage. It is compliant with all standards for microarray meta-data capture; including the specification of what data should be recorded, extensive use of standard ontologies and support for data exchange formats. The output from maxdLoad2 is of a form acceptable for submission to the ArrayExpress microarray repository at the European Bioinformatics Institute. maxdBrowse is a PHP web-application that makes contents of maxdLoad2 databases accessible via web-browser, the command-line and web-service environments. It thus acts as both a dissemination and data-mining tool.ResultsmaxdLoad2 presents an easy-to-use interface to an underlying relational database and provides a full complement of facilities for browsing, searching and editing. There is a tree-based visualization of data connectivity and the ability to explore the links between any pair of data elements, irrespective of how many intermediate links lie between them. Its principle novel features are:• the flexibility of the meta-data that can be captured,• the tools provided for importing data from spreadsheets and other tabular representations,• the tools provided for the automatic creation of structured documents,• the ability to browse and access the data via web and web-services interfaces.Within maxdLoad2 it is very straightforward to customise the meta-data that is being captured or change the definitions of the meta-data. These meta-data definitions are stored within the database itself allowing client software to connect properly to a modified database without having to be specially configured. The meta-data definitions (configuration file) can also be centralized allowing changes made in response to revisions of standards or terminologies to be propagated to clients without user intervention.maxdBrowse is hosted on a web-server and presents multiple interfaces to the contents of maxd databases. maxdBrowse emulates many of the browse and search features available in the maxdLoad2 application via a web-browser. This allows users who are not familiar with maxdLoad2 to browse and export microarray data from the database for their own analysis. The same browse and search features are also available via command-line and SOAP server interfaces. This both enables scripting of data export for use embedded in data repositories and analysis environments, and allows access to the maxd databases via web-service architectures.ConclusionmaxdLoad2 and maxdBrowse are portable and compatible with all common operating systems and major database servers. They provide a powerful, flexible package for annotation of microarray experiments and a convenient dissemination environment. They are available for download and open sourced under the Artistic License.