Bayesian Learning with Dependency Structures via Latent Factors, Mixtures, and Copulas

Bayesian methods offer a flexible and convenient probabilistic learning framework to extract interpretable knowledge from complex and structured data. Such methods can characterize dependencies among multiple levels of hidden variables and share statistical strength across heterogeneous sources. In the first part of this dissertation, we develop two dependent variational inference methods for full posterior approximation in non-conjugate Bayesian models through hierarchical mixture- and copula-based variational proposals, respectively. The proposed methods move beyond the widely used factorized approximation to the posterior and provide generic applicability to a broad class of probabilistic models with minimal model-specific derivations. In the second part of this dissertation, we design probabilistic graphical models to accommodate multimodal data, describe dynamical behaviors and account for task heterogeneity. In particular, the sparse latent factor model is able to reveal common low-dimensional structures from high-dimensional data. We demonstrate the effectiveness of the proposed statistical learning methods on both synthetic and real-world data.

(Table 1) Characteristics of marine snow and ice samples at Barrow, Alaska

We measured light absorption in 42 marine snow, sea ice, seawater, brine, and frost flower samples collected during the OASIS field campaign between February 27 and April 15, 2009. Samples represented multiple sites between landfast ice and open pack ice in coastal areas approximately 5 km west of Barrow, Alaska. The chromophores that are most commonly measured in snow, H2O2, NO3-, and NO2-, on average account for less than 1% of sunlight absorption in our samples. Instead, light absorption is dominated by unidentified "residual" species, likely organic compounds. Light absorption coefficients for the frost flowers on first-year sea ice are, on average, 40 times larger than values for terrestrial snow samples at Barrow, suggesting very large rates of photochemical reactions in frost flowers. For our marine samples the calculated rates of sunlight absorption and OH production from known chromophores are (0.1-1.4) x 10**14 (photons/cm**3/s) and (5-70) x 10**-12 (mol/L/s), respectively. Our residual spectra are similar to spectra of marine chromophoric dissolved organic matter (CDOM), suggesting that CDOM is the dominant chromophore in our samples. Based on our light absorption measurements we estimate dissolved organic carbon (DOC) concentrations in Barrow seawater and frost flowers as approximately 130 and 360 µM C, respectively. We expect that CDOM is a major source of OH in our marine samples, and it is likely to have other significant photochemistry as well.

The Rotational Evolution and Magnetospheric Emission of the Magnetic Early B-type Stars

How do the magnetic fields of massive stars evolve over time? Are their gyrochronological ages consistent with ages inferred from evolutionary tracks? Why do most stars predicted to host Centrifugal Magnetospheres (CMs) display no H$\alpha$ emission? Does plasma escape from CMs via centrifugal breakout events, or by a steady-state leakage mechanism? This thesis investigates these questions via a population study with a sample of 51 magnetic early B-type stars. The longitudinal magnetic field \bz~was measured from Least Squares Deconvolution profiles extracted from high-resolution spectropolarimetric data. New rotational periods $P_{\rm rot}$ were determined for 15 stars from \bz, leaving only 3 stars for which $P_{\rm rot}$ is unknown. Projected rotational velocities \vsini~were measured from multiple spectral lines. Effective temperatures and surface gravities were measured via ionization balances and line profile fitting of H Balmer lines. Fundamental physical parameters, \bz, \vsini, and $P_{\rm rot}$ were then used to determine radii, masses, ages, dipole oblique rotator model, stellar wind, magnetospheric, and spindown parameters using a Monte Carlo approach that self-consistently calculates all parameters while accounting for all available constraints on stellar properties. Dipole magnetic field strengths $B_{\rm d}$ follow a log-normal distribution similar to that of Ap stars, and decline over time in a fashion consistent with the expected conservation of fossil magnetic flux. $P_{\rm rot}$ increases with fractional main sequence age, mass, and $B_{\rm d}$, as expected from magnetospheric braking. However, comparison of evolutionary track ages to maximum spindown ages $t_{\rm S,max}$ shows that initial rotation fractions may be far below critical for stars with $M_*>10 M_\odot$. Computing $t_{\rm S,max}$ with different mass-loss prescriptions indicates that the mass-loss rates of B-type stars are likely much lower than expected from extrapolation from O-type stars. Stars with H$\alpha$ in emission and absorption occupy distinct regions in the updated rotation-magnetic confinement diagram: H$\alpha$-bright stars are found to be younger, more rapidly rotating, and more strongly magnetized than the general population. Emission strength is sensitive both to the volume of the CM and to the mass-loss rate, favouring leakage over centrifugal breakout.

Supporting Metacognitive Development in Early Science Education: Exploring Elementary Teachers' Beliefs and Practices in Metacognition

Metacognition is the understanding and control of cognitive processes. Students with high levels of metacognition achieve greater academic success. The purpose of this mixed-methods study was to examine elementary teachers’ beliefs about metacognition and integration of metacognitive practices in science. Forty-four teachers were recruited through professional networks to complete a questionnaire containing open-ended questions (n = 44) and Likert-type items (n = 41). Five respondents were selected to complete semi-structured interviews informed by the questionnaire. The selected interview participants had a minimum of three years teaching experience and demonstrated a conceptual understanding of metacognition. Statistical tests (Pearson correlation, t-tests, and multiple regression) on quantitative data and thematic analysis of qualitative data indicated that teachers largely understood metacognition but had some gaps in their understanding. Participants’ reported actions (teaching practices) and beliefs differed according to their years of experience but not gender. Hierarchical multiple regression demonstrated that the first block of gender and experience was not a significant predictor of teachers' metacognitive actions, although experience was a significant predictor by itself. Experience was not a significant predictor once teachers' beliefs were added. The majority of participants indicated that metacognition was indeed appropriate for elementary students. Participants consistently reiterated that students’ metacognition developed with practice, but required explicit instruction. A lack of consensus remained around the domain specificity of metacognition. More specifically, the majority of questionnaire respondents indicated that metacognitive strategies could not be used across subject domains, whereas all interviewees indicated that they used strategies across subjects. Metacognition was integrated frequently into Ontario elementary classrooms; however, metacognition was integrated less frequently in science lessons. Lastly, participants used a variety of techniques to integrate metacognition into their classrooms. Implications for practice include the need for more professional development aimed at integrating metacognition into science lessons at both the Primary and Junior levels. Further, teachers could benefit from additional clarification on the three main components of metacognition and the need to integrate all three to successfully develop students’ metacognition.

Partner selection for reverse logistics centres in green supply chains: a fuzzy artificial immune optimization approach

The design of reverse logistics networks has now emerged as a major issue for manufacturers, not only in developed countries where legislation and societal pressures are strong, but also in developing countries where the adoption of reverse logistics practices may offer a competitive advantage. This paper presents a new model for partner selection for reverse logistic centres in green supply chains. The model offers three advantages. Firstly, it enables economic, environment, and social factors to be considered simultaneously. Secondly, by integrating fuzzy set theory and artificial immune optimization technology, it enables both quantitative and qualitative criteria to be considered simultaneously throughout the whole decision-making process. Thirdly, it extends the flat criteria structure for partner selection evaluation for reverse logistics centres to the more suitable hierarchy structure. The applicability of the model is demonstrated by means of an empirical application based on data from a Chinese electronic equipment and instruments manufacturing company.

The Narrative Value of Dance Communication: "habitus", musicality and emotion

The richness of dance comes from the need to work with an individual body. Still, the body of the dancer belongs to plural context, crossed by artistic and social traditions, which locate the artists in a given field. We claim that role conflict is an essential component of the structure of collective artistic creativity. We address the production of discourse in a British dance company, with data that spawns from the ethnography ‘Dance and Cognition’, directed by David Kirsh at the University of California, together with WayneMcGregor-Random Dance. Our Critical Discourse Analysis is based on multiple interviews to the dancers and choreographer. Our findings show how creativity in dance seems to be empirically observable, and thus embodied and distributed shaped by the dance habitus of the particular social context.

"Sub Tiberio quies"? Judaea’s Political Situation under the Prefects (6-41 c.e.), between Reality and Ideology

Deeply conflicting views on the political situation of Judaea under the Roman prefects (6-41 c.e.) have been offered. According to some scholars, this was a period of persistent political unrest and agitation, whilst according to a widespread view it was a quiescent period of political calm (reflected in Tacitus’ phrase sub Tiberio quies). The present article critically examines again the main available sources –particularly Josephus, the canonical Gospels and Tacitus– in order to offer a more reliable historical reconstruction. The conclusions drawn by this survey calls into question some widespread and insufficiently nuanced views on the period. This, in turn, allows a reflection on the non-epistemic factors which might contribute to explain the origin of such views.

La Ciberseguridad como factor crítico en la Seguridad de la Unión Europea

El ciberespacio es un escenario de conflicto altamente complejo al estar en constante evolución. Ni la Unión Europea ni ningún otro actor del sistema internacional se encuentra a salvo de las amenazas procedentes del ciberespacio. Pero los pasos dados desde la UE en el mundo de la ciberseguridad no son en absoluto suficientes. Europa necesita que su Estrategia de ciberseguridad sea realmente capaz de integrar a las diferentes Estrategias nacionales. Es urgente una mayor determinación, unos mayores recursos y unos mejores instrumentos que permitan a la Unión implementar una gestión de crisis y una prevención de ciberconflictos verdaderamente eficaz.

Withdrawal of Antidementia Drugs in Older People: Who, When and How?

The evidence base to guide withdrawal of antidementia medications in older people with dementia is limited; while some randomised controlled studies have considered discontinuation of cholinesterase inhibitors, no such studies examining discontinuation of the N-Methyl-D-aspartate receptor antagonist memantine have been conducted to date. The purpose of this opinion article was to summarise the existing evidence on withdrawal of cholinesterase inhibitors and memantine, to highlight the key considerations for clinicians when making these prescribing decisions and to offer guidance as to when and how treatment might be discontinued. Until the evidence-base is enhanced by the findings of large scale randomised controlled discontinuation trials of ChEIs and memantine which use multiple, clinically relevant cognitive, functional and behavioural outcome measures, clinicians’ prescribing decisions involve balancing the risks of discontinuation with side-effects and costs of continued treatment. Such decisions must be highly individualised and patient-centred.

A Generalized Fuzzy Linguistic Model for Predicting Component Concentrations in an Optical Gas Sensing System

Motivated by environmental protection concerns, monitoring the flue gas of thermal power plant is now often mandatory due to the need to ensure that emission levels stay within safe limits. Optical based gas sensing systems are increasingly employed for this purpose, with regression techniques used to relate gas optical absorption spectra to the concentrations of specific gas components of interest (NOx, SO2 etc.). Accurately predicting gas concentrations from absorption spectra remains a challenging problem due to the presence of nonlinearities in the relationships and the high-dimensional and correlated nature of the spectral data. This article proposes a generalized fuzzy linguistic model (GFLM) to address this challenge. The GFLM is made up of a series of “If-Then” fuzzy rules. The absorption spectra are input variables in the rule antecedent. The rule consequent is a general nonlinear polynomial function of the absorption spectra. Model parameters are estimated using least squares and gradient descent optimization algorithms. The performance of GFLM is compared with other traditional prediction models, such as partial least squares, support vector machines, multilayer perceptron neural networks and radial basis function networks, for two real flue gas spectral datasets: one from a coal-fired power plant and one from a gas-fired power plant. The experimental results show that the generalized fuzzy linguistic model has good predictive ability, and is competitive with alternative approaches, while having the added advantage of providing an interpretable model.

Fuzzy logic-based implicit authentication for mobile access control

In order to address the increasing compromise of user privacy on mobile devices, a Fuzzy Logic based implicit authentication scheme is proposed in this paper. The proposed scheme computes an aggregate score based on selected features and a threshold in real-time based on current and historic data depicting user routine. The tuned fuzzy system is then applied to the aggregated score and the threshold to determine the trust level of the current user. The proposed fuzzy-integrated implicit authentication scheme is designed to: operate adaptively and completely in the background, require minimal training period, enable high system accuracy while provide timely detection of abnormal activity. In this paper, we explore Fuzzy Logic based authentication in depth. Gaussian and triangle-based membership functions are investigated and compared using real data over several weeks from different Android phone users. The presented results show that our proposed Fuzzy Logic approach is a highly effective, and viable scheme for lightweight real-time implicit authentication on mobile devices.

A generalized fuzzy linguistic model for predicting component concentrations in an optical gas sensing system

Motivated by environmental protection concerns, monitoring the flue gas of thermal power plant is now often mandatory due to the need to ensure that emission levels stay within safe limits. Optical based gas sensing systems are increasingly employed for this purpose, with regression techniques used to relate gas optical absorption spectra to the concentrations of specific gas components of interest (NO_x, SO₂ etc.). Accurately predicting gas concentrations from absorption spectra remains a challenging problem due to the presence of nonlinearities in the relationships and the high-dimensional and correlated nature of the spectral data. This article proposes a generalized fuzzy linguistic model (GFLM) to address this challenge. The GFLM is made up of a series of “If-Then” fuzzy rules. The absorption spectra are input variables in the rule antecedent. The rule consequent is a general nonlinear polynomial function of the absorption spectra. Model parameters are estimated using least squares and gradient descent optimization algorithms. The performance of GFLM is compared with other traditional prediction models, such as partial least squares, support vector machines, multilayer perceptron neural networks and radial basis function networks, for two real flue gas spectral datasets: one from a coal-fired power plant and one from a gas-fired power plant. The experimental results show that the generalized fuzzy linguistic model has good predictive ability, and is competitive with alternative approaches, while having the added advantage of providing an interpretable model.

Boys2Men, an age estimation dataset with applications to detect enfants in pornography content

[EN]The aim of this paper is the detection of non adults in images.

Methods, Models & Systems for Dealing with Some Business, Management & Engineering Problems

This keynote presentation will report some of our research work and experience on the development and applications of relevant methods, models, systems and simulation techniques in support of different types and various levels of decision making for business, management and engineering. In particular, the following topics will be covered. Modelling, multi-agent-based simulation and analysis of the allocation management of carbon dioxide emission permits in China (Nanfeng Liu & Shuliang Li Agent-based simulation of the dynamic evolution of enterprise carbon assets (Yin Zeng & Shuliang Li) A framework & system for extracting and representing project knowledge contexts using topic models and dynamic knowledge maps: a big data perspective (Jin Xu, Zheng Li, Shuliang Li & Yanyan Zhang) Open innovation: intelligent model, social media & complex adaptive system simulation (Shuliang Li & Jim Zheng Li) A framework, model and software prototype for modelling and simulation for deshopping behaviour and how companies respond (Shawkat Rahman & Shuliang Li) Integrating multiple agents, simulation, knowledge bases and fuzzy logic for international marketing decision making (Shuliang Li & Jim Zheng Li) A Web-based hybrid intelligent system for combined conventional, digital, mobile, social media and mobile marketing strategy formulation (Shuliang Li & Jim Zheng Li) A hybrid intelligent model for Web & social media dynamics, and evolutionary and adaptive branding (Shuliang Li) A hybrid paradigm for modelling, simulation and analysis of brand virality in social media (Shuliang Li & Jim Zheng Li) Network configuration management: attack paradigms and architectures for computer network survivability (Tero Karvinen & Shuliang Li)

Context, Complexity and Contestation: Birmingham’s Agreed Syllabuses for Religious Education since the 1970s

The present article offers an historical perspective on the 1975, 1995 and 2007 Birmingham Agreed Syllabuses for Religious Education. It draws upon historical evidence uncovered as part of ‘The hidden history of curriculum change in reli- gious education in English schools, 1969–1979’ project, and curriculum history theories, especially David Labaree’s observations about the distance between the ‘rhetorical’ and ‘received’ curricula. We argue that, contrary to the existing his- toriography, curriculum change in religious education (RE) has been evolution- ary not revolutionary. Multiple reasons are posited to explain this, not least among which is the capacity and agency of teachers. Furthermore, we argue that ongoing debates about the nature and purpose of RE, as exemplified in the Birmingham context, reflect the multiple expectations that religious educators and other stakeholders had, and continue to have, of the curriculum subject. These debates contribute to the inertia evident in the implementation of RE cur- riculum reforms. A consciousness of the history of RE enables curriculum con- testations to be contextualised and understood, and, thereby, provides important insights which can be applied to ongoing and future debates and developments.