Learning a Family of Detectors

Object detection and recognition are important problems in computer vision. The challenges of these problems come from the presence of noise, background clutter, large within class variations of the object class and limited training data. In addition, the computational complexity in the recognition process is also a concern in practice. In this thesis, we propose one approach to handle the problem of detecting an object class that exhibits large within-class variations, and a second approach to speed up the classification processes. In the first approach, we show that foreground-background classification (detection) and within-class classification of the foreground class (pose estimation) can be jointly solved with using a multiplicative form of two kernel functions. One kernel measures similarity for foreground-background classification. The other kernel accounts for latent factors that control within-class variation and implicitly enables feature sharing among foreground training samples. For applications where explicit parameterization of the within-class states is unavailable, a nonparametric formulation of the kernel can be constructed with a proper foreground distance/similarity measure. Detector training is accomplished via standard Support Vector Machine learning. The resulting detectors are tuned to specific variations in the foreground class. They also serve to evaluate hypotheses of the foreground state. When the image masks for foreground objects are provided in training, the detectors can also produce object segmentation. Methods for generating a representative sample set of detectors are proposed that can enable efficient detection and tracking. In addition, because individual detectors verify hypotheses of foreground state, they can also be incorporated in a tracking-by-detection frame work to recover foreground state in image sequences. To run the detectors efficiently at the online stage, an input-sensitive speedup strategy is proposed to select the most relevant detectors quickly. The proposed approach is tested on data sets of human hands, vehicles and human faces. On all data sets, the proposed approach achieves improved detection accuracy over the best competing approaches. In the second part of the thesis, we formulate a filter-and-refine scheme to speed up recognition processes. The binary outputs of the weak classifiers in a boosted detector are used to identify a small number of candidate foreground state hypotheses quickly via Hamming distance or weighted Hamming distance. The approach is evaluated in three applications: face recognition on the face recognition grand challenge version 2 data set, hand shape detection and parameter estimation on a hand data set, and vehicle detection and estimation of the view angle on a multi-pose vehicle data set. On all data sets, our approach is at least five times faster than simply evaluating all foreground state hypotheses with virtually no loss in classification accuracy.

Molecular analysis of virulence mechanisms associated with adherent-invasive Escherichia coli (AIEC)

Crohn's Disease (CD) is a chronic inflammatory bowel disease of unknown etiology. Recent work has shown that a new pathotype of Escherichia coli, Adherent Invasive E. coli (AIEC) may be associated with CD. AIEC has been shown to adhere to and invade epithelial cells and to replicate within macrophages (together this is called the AIEC phenotype). In this thesis, the AIEC phenotype of 84 E. coli strains were determined in order to identify the prevalence of this phenotype within the E. coli genus. This study showed that a significant proportion of E. coli strains (approx. 5%) are capable of adhering to and invading epithelial cells and undergoing intramacrophage replication. Moreover, the results presented in this study indicate a correlation between survival in macrophage and resistance to grazing by amoeba supporting the coincidental evolution hypothesis that resistance to amoebae could be a driving force in the evolution of pathogenicity in some bacteria, such as AIEC. In addition, this study has identified an important regulatory role for the CpxA/R two component system (TCS) in the invasive abilities of AIEC HM605, a colonic mucosa-associated CD isolate. A mutation in cpxR was shown to be defective in the invasion of epithelial cells and this defect was shown to be independent of motility or the expression of Type 1 fimbriae, factors that have been shown to be involved in the invasion of another strain of AIEC, isolated from a patient with ileal CD, called LF82. The CpxA/R TCS responds to disturbances in the cell envelope and has been implicated in the virulence of a number of Gram negative pathogens. In this study it is shown that the CpxA/R TCS regulates the expression of a potentially novel invasin called SinH. SinH is found in a number of invasive strains of E. coli and Salmonella. Moreover work presented here shows that a critical mechanism underpinning AIEC persistence in macrophages is the repair of DNA bases damaged by macrophage oxidants. Together these findings provide evidence to suggest that AIEC are a diverse group of E. coli and possess diverse molecular mechanisms and virulence factors that contribute to the AIEC phenotype. In addition, AIEC may have gone through different evolutionary histories acquiring various molecular mechanisms ultimately culminating in the AIEC phenotype. The gastrointestinal (GI) tract harbors a diverse microbiota; most are symbiotic or commensal however some bacteria have the potential to cause disease (pathobiont). The work presented here provides evidence to support the model that AIEC are pathobionts. AIEC strains can be carried as commensals in healthy guts however, when the intestinal homeostasis is disrupted, such as in the compromised gut of CD patients, AIEC may behave as opportunistic pathogens and cause and/or contribute to disease by driving intestinal inflammation.

Developing executive capability — banking and beyond

The transition to becoming a leader is perhaps the least understood and most difficult in business. This Portfolio of Exploration examines the development of conscious awareness and meaning complexity as key transformational requirements to operate competently at leadership level and to succeed in a work environment characterised by change and complexity. It recognises that developing executive leadership capability is not just an issue of personality increasing what we know or expertise. It requires development of complexity in terms of how we know ourselves, relate to others, construe leadership and organisation, problem solve in business and understand the world as a whole. The exploration is grounded in the theory of adult mental development as outlined by Robert Kegan (1982, 1994) and in his collaborations with Lisa Laskow Lahey (2001, 2009). The theory points to levels of consciousness which impact on how we make meaning of and experience the world around us and respond to it. Critically it also points to transformational processes which enable us to evolve how we make meaning of our world as a means to close the mismatch between the demands of this world and our ability to cope. The exploration is laid out in three stages. Using Kegan’s (1982, 1994) theory as a framework it begins with a reflection of my career to surface how I made meaning of banking, management and subsequently leadership. In stage two I engage with a range of source thinkers in the areas of leadership, decision making, business, organisation, growth and complexity in a transformational process of developing greater conscious and complex understanding of organisational leadership (also recognising ever increasing complexity in the world). Finally, in stage three, I explore how qualitative changes as a result of this transformational effort have benefitted my professional, leadership and organisational capabilities.

The vertebrate taxonomy ontology: a framework for reasoning across model organism and species phenotypes.

BACKGROUND: A hierarchical taxonomy of organisms is a prerequisite for semantic integration of biodiversity data. Ideally, there would be a single, expansive, authoritative taxonomy that includes extinct and extant taxa, information on synonyms and common names, and monophyletic supraspecific taxa that reflect our current understanding of phylogenetic relationships. DESCRIPTION: As a step towards development of such a resource, and to enable large-scale integration of phenotypic data across vertebrates, we created the Vertebrate Taxonomy Ontology (VTO), a semantically defined taxonomic resource derived from the integration of existing taxonomic compilations, and freely distributed under a Creative Commons Zero (CC0) public domain waiver. The VTO includes both extant and extinct vertebrates and currently contains 106,947 taxonomic terms, 22 taxonomic ranks, 104,736 synonyms, and 162,400 cross-references to other taxonomic resources. Key challenges in constructing the VTO included (1) extracting and merging names, synonyms, and identifiers from heterogeneous sources; (2) structuring hierarchies of terms based on evolutionary relationships and the principle of monophyly; and (3) automating this process as much as possible to accommodate updates in source taxonomies. CONCLUSIONS: The VTO is the primary source of taxonomic information used by the Phenoscape Knowledgebase (http://phenoscape.org/), which integrates genetic and evolutionary phenotype data across both model and non-model vertebrates. The VTO is useful for inferring phenotypic changes on the vertebrate tree of life, which enables queries for candidate genes for various episodes in vertebrate evolution.

Efects of mineral suspension and dissolution on strength and compressibility of soft carbonate rocks

© 2014 Elsevier B.V.Calcarenites are highly porous soft rocks formed of mainly carbonate grains bonded together by calcite bridges. The above characteristics make them prone to water-induced weathering, frequently featuring large caverns and inland natural underground cavities. This study is aimed to determine the main physical processes at the base of the short- and long-term weakening experienced by these rocks when interacting with water. We present the results of microscale experimental investigations performed on calcarenites from four different sites in Southern Italy. SEM, thin sections, X-ray CT observations and related analyses are used for both the interpretation-definition of the structure changes, and the identification-quantification of the degradation mechanisms. Two distinct types of bonding have been identified within the rock: temporary bonding (TB) and persistent bonding (PB). The diverse mechanisms linked to these two types of bonding explain both the observed fast decrease in rock strength when water fills the pores (short-term effect of water), identified with a short-term debonding (STD), and a long-term weakening of the material, when the latter is persistently kept in water-saturated conditions (long-term effect of water), identified with a long-term debonding (LTD). To highlight the micro-hydro-chemo-mechanical processes of formation and annihilation of the TB bonds and their role in the evolution of the mechanical strength of the material, mechanical tests on samples prepared by drying partially saturated calcarenite powder, or a mix of glass ballotini and calcarenite powder were conducted. The long-term debonding processes have also been investigated, using acid solutions in order to accelerate the reaction rates. This paper attempts to identify and quantify differences between the two types of bonds and the relative micro-scale debonding processes leading to the macro-scale material weakening mechanisms.

A robust tri-state election algorithm for scalable applications

Existing election algorithms suffer limited scalability. This limit stems from the communication design which in turn stems from their fundamentally two-state behaviour. This paper presents a new election algorithm specifically designed to be highly scalable in broadcast networks whilst allowing any processing node to become coordinator with initially equal probability. To achieve this, careful attention has been paid to the communication design, and an additional state has been introduced. The design of the tri-state election algorithm has been motivated by the requirements analysis of a major research project to deliver robust scalable distributed applications, including load sharing, in hostile computing environments in which it is common for processing nodes to be rebooted frequently without notice. The new election algorithm is based in-part on a simple 'emergent' design. The science of emergence is of great relevance to developers of distributed applications because it describes how higher-level self-regulatory behaviour can arise from many participants following a small set of simple rules. The tri-state election algorithm is shown to have very low communication complexity in which the number of messages generated remains loosely-bounded regardless of scale for large systems; is highly scalable because nodes in the idle state do not transmit any messages; and because of its self-organising characteristics, is very stable.

AC and DC magnetic levitation: melting, fluid flow and oscillations

The presented numerical modelling for the magnetic levitation involves coupling of the electromagnetic field, liquid shape change, fluid velocities and the temperature field at every time step during the simulation in time evolution. Combination of the AC and DC magnetic fields can be used to achieve high temperature, stable levitation conditions. The oscillation frequency spectra are analysed for droplets levitated in AC and DC magnetic fields at various combinations. An electrically poorly conducting, diamagnetic droplet (e.g. water) can be stably levitated using the dia- and para-magnetic properties of the sample material in a high intensity, gradient DC field.

Three-dimensional simulation of the influence of convection on dendritic solidification

Bulk and interdendritic flow during solidification alters the microstructure development, potentially leading to the formation of defects. In this paper, a 3D numerical model is presented for the simulation of dendritic growth in the presence of fluid flow in both liquid and semi-solid zones during solidification. The dendritic growth was solved by the combination of a stochastic nucleation approach with a finite difference solution of the solute diffusion equation and. a projection method solution of the Navier-Stokes equations. The technique was applied first to simulate the growth of a single dendrite in 2D and 3D in an isothermal environment with forced fluid flow. Significant differences were found in the evolution of dendritic morphology when comparing the 2D and 3D results. In 3D the upstream arm has a faster growth velocity due to easier flow around the perpendicular arms. This also promotes secondary arm formation on the upstream arm. The effect of fluid flow on columnar dendritic growth and micro-segregation in constrained solidification conditions is then simulated. For constrained growth, 2D simulations lead to even greater inaccuracies as compared to 3D.

Sardine cycles, krill declines, and locust plagues: revisiting ‘wasp-waist’ food webs

‘Wasp-waist’ systems are dominated by a mid trophic-level species that is thought to exert top-down control on its food and bottom-up control on its predators. Sardines, anchovy, and Antarctic krill are suggested examples, and here we use locusts to explore whether the wasp-waist concept also applies on land. These examples also display the traits of mobile aggregations and dietary diversity, which help to reduce the foraging footprint from their large, localised biomasses. This suggests that top-down control on their food operates at local aggregation scales and not at wider scales suggested by the original definition of wasp-waist. With this modification, the wasp-waist framework can cross-fertilise marine and terrestrial approaches, revealing how seemingly disparate but economically important systems operate.

High-performance 3D median filter architecture for medical image despeckling

A high-sample rate 3D median filtering processor architecture is proposed, based on a novel 3D median filtering algorithm, that can reduce the computing complexity in comparison with the traditional bubble sorting algorithm. A 3 x 3 x 3 filter processor is implemented in VHDL, and the simulation verifies that the processor can process a 128 x 128 x 96 MRI image in 0.03 seconds while running at 50 MHz.

Phylogeographic insights into cryptic glacial refugia

The glacial episodes of the Quaternary (2.6 million years ago–present) were a major factor in shaping the present-day distributions of extant flora and fauna, with expansions and contractions of the ice sheets rendering large areas uninhabitable for most species. Fossil records suggest that many species survived glacial maxima by retreating to refugia, usually at lower latitudes. Recently, phylogeographic studies have given support to the existence of previously unknown, or cryptic, refugia. Here we summarise many of these insights into the glacial histories of species in cryptic refugia gained through phylogeographic approaches. Understanding such refugia might be important as the Earth heads into another period of climate change, in terms of predicting the effects on species distribution and survival.

How parasites affect interactions between competitors and predators

We present a synthesis of empirical and theoretical work investigating how parasites influence competitive and predatory interactions between other species. We examine the direct and indirect effects of parasitism and discuss examples of density and parasite-induced trait-mediated effects. Recent work reveals previously unrecognized complexity in parasite-mediated interactions. In addition to parasite-modified and apparent competition leading to species exclusion or enabling coexistence, parasites and predators interact in different ways to regulate or destablize the population dynamics of their joint prey. An emerging area is the impact of parasites on intraguild predation (IGP). Parasites can increase vulnerability of infected individuals to cannibalism or predation resulting in reversed species dominance in IGP hierarchies. We discuss the potential significance of parasites for community structure and biodiversity, in particular their role in promoting species exclusion or coexistence and the impact of emerging diseases. Ongoing invasions provide examples where parasites mediate native/invader interactions and play a key role in determining the outcome of invasions. We highlight the need for more quantitative data to assess the impact of parasites on communities, and the combination of theoretical and empirical studies to examine how the effects of parasitism scale up to community-level processes.

Tracing early stages of species differentiation: Ecological, morphological and genetic divergence of Galapagos sea lion populations

Background: Oceans are high gene flow environments that are traditionally believed to hamper the build-up of genetic divergence. Despite this, divergence appears to occur occasionally at surprisingly small scales. The Galápagos archipelago provides an ideal opportunity to examine the evolutionary processes of local divergence in an isolated marine environment. Galápagos sea lions (Zalophus wollebaeki) are top predators in this unique setting and have an essentially unlimited dispersal capacity across the entire species range. In theory, this should oppose any genetic differentiation.
Results: We find significant ecological, morphological and genetic divergence between the western colonies and colonies from the central region of the archipelago that are exposed to different ecological conditions. Stable isotope analyses indicate that western animals use different food sources than those from the central area. This is likely due to niche partitioning with the second Galápagos eared seal species, the Galápagos fur seal (Arctocephalus galapagoensis) that exclusively dwells in the west. Stable isotope patterns correlate with significant differences in foraging-related skull morphology. Analyses of mitochondrial sequences as well as microsatellites reveal signs of initial genetic differentiation.
Conclusion: Our results suggest a key role of intra- as well as inter-specific niche segregation in the evolution of genetic structure among populations of a highly mobile species under conditions of free movement. Given the monophyletic arrival of the sea lions on the archipelago, our study challenges the view that geographical barriers are strictly needed for the build-up of genetic divergence. The study further raises the interesting prospect that in social, colonially breeding mammals additional forces, such as social structure or feeding traditions, might bear on the genetic partitioning of populations.

Doubly duped males: the sweet and sour of the orchid's bouquet

Flowers of the orchid genus Ophrys resemble female insects, and thereby sexually deceive, attract and are pollinated by male insects. Floral bouquet is thought to play a major role in this sexual mimicry, although the search for functional odour components has been something of a chemical ecologist's Holy Grail. Two new papers unravel the exquisite intricacy of the chemical deception by the orchid.

Novel periodically loaded E-plane filters

Novel E-plane waveguide filters with periodically loaded resonators are proposed. The proposed filters make use of the slow wave effect in order to achieve improved stopband performance and size reduction of roughly 50% without introducing any complexity in the fabrication process. Numerical and experimental results are presented to validate the argument.