Studies on the generalisation of Gaussian processes and Bayesian neural networks

The assessment of the reliability of systems which learn from data is a key issue to investigate thoroughly before the actual application of information processing techniques to real-world problems. Over the recent years Gaussian processes and Bayesian neural networks have come to the fore and in this thesis their generalisation capabilities are analysed from theoretical and empirical perspectives. Upper and lower bounds on the learning curve of Gaussian processes are investigated in order to estimate the amount of data required to guarantee a certain level of generalisation performance. In this thesis we analyse the effects on the bounds and the learning curve induced by the smoothness of stochastic processes described by four different covariance functions. We also explain the early, linearly-decreasing behaviour of the curves and we investigate the asymptotic behaviour of the upper bounds. The effect of the noise and the characteristic lengthscale of the stochastic process on the tightness of the bounds are also discussed. The analysis is supported by several numerical simulations. The generalisation error of a Gaussian process is affected by the dimension of the input vector and may be decreased by input-variable reduction techniques. In conventional approaches to Gaussian process regression, the positive definite matrix estimating the distance between input points is often taken diagonal. In this thesis we show that a general distance matrix is able to estimate the effective dimensionality of the regression problem as well as to discover the linear transformation from the manifest variables to the hidden-feature space, with a significant reduction of the input dimension. Numerical simulations confirm the significant superiority of the general distance matrix with respect to the diagonal one.In the thesis we also present an empirical investigation of the generalisation errors of neural networks trained by two Bayesian algorithms, the Markov Chain Monte Carlo method and the evidence framework; the neural networks have been trained on the task of labelling segmented outdoor images.

Influence of iron deprivation and sub-inhibitory concentrations of antifungal antibiotics on surface antigens of candida albicans yeast cells

This study examined the effect of iron deprivation and sub-inhibitory concentrations of antifungal agents on yeast cell surface antigen recognition by antibodies from patients with Candida infections. Separation of cell wall surface proteins by sodium dodecyl-polyacrylamide gel electrophoresis (SDS-PAGE) and immunological detection by immunoblotting, revealed that antigenic profiles of yeasts were profoundly influenced by the growth environment. Cells grown under iron-depleted conditions expressed several iron-regulated proteins that were recognized by antibodies from patient sera. An attempt to characterize these proteins by lectin blotting with concanavalin A revealed that some could be glycoprotein in nature. Furthermore, these proteins which were located within cell walls and on yeast surfaces, were barely or not expressed in yeasts cultivated under iron-sufficient conditions. The magnitude and heterogeneity of human antibody responses to these iron-regulated proteins were dependent on the type of Candida infection, serum antibody class and yeast strain. Hydroxamate-type siderophores were also detected in supernatants of iron depleted yeast cultures. This evidence suggests that Candida albicans expresses iron-regulated proteins/glycoproteins in vitro which may play a role in siderophore-mediated iron uptake in Candida albicans. Sequential monitoring of IgG antibodies directed against yeast surface antigens during immunization of rabbits revealed that different antigens were recognized particularly during early and later stages of immunization in iron-depleted cells compared to iron-sufficient cells. In vitro and in vivo adherence studies demonstrated that growth phase, yeast strain and growth conditions affect adhesion mechanisms. In particular, growth under iron-depletion in the presence of sub-inhibitory concentrations of polyene and azole antifungals enhanced the hydrophobicity of C.albicans. Growth conditions also influenced MICs of antifungals, notably that of ketoconazole. Sub-inhibitory concentrations of amphotericin B and fluconazole had little effect on surface antigens, whereas nystatin induced profound changes in surface antigens of yeast cells. The effects of such drug concentrations on yeast cells coupled with host defence mechanisms may have a significant affect on the course of Candida infections.

Consumer responses to really new products: examining the impacts of learning strategies and presentation formats on product comprehension and attitude formation

Mental simulations and analogies have been identified as powerful learning tools for RNPs. Furthermore, visuals in advertising have recently been conceptualized as meaningful sources of information as opposed to peripheral cues and thus may help consumers learn about RNPs. The study of visual attention may also contribute to understanding the links between conceptual and perceptual analyses when learning for a RNP. Two conceptual models are developed. the first model consists of causal relationships between the attributes of advertising stimuli for RNPs and consumer responses, as well as mediating influences. The second model focuses on the role of visual attention in product comprehension as a response to advertising stimuli. Two experiments are conducted: a Web-Experiment and an eye-tracking experiment. The first experiment (858 subjects) examines the effect of learning strategies (mental simulation vs. analogy vs. no analogy/no mental simulation) and presentation formats (words vs. pictures) on individual responses. The mediating role of emotions is assessed. The second experiment investigates the effect of learning strategies and presentation formats on product comprehension, along with the role of attention (17 subjects). The findings from experiment 1 indicate that learning strategies and presentation formats can either enhance or undermine the effect of advertising stimuli on individual responses. Moreover, the nature of the product (i.e. hedonic vs. utilitarian vs. hybrid) should be considered when designing communications for RNPs. The mediating role of emotions is verified. Experiment 2 suggests that an increase in attention to the message may either reflect enhanced comprehension or confusion.

The effect of innate compliance on the performance of a counterpulsation device

Cardiovascular disease (CVD) continues to be one of the top causes of mortality in the world. World Heart Organization (WHO) reported that in 2004, CVD contributed to almost 30% of death from estimated worldwide death figures of 58 million[1]. Heart failure treatment varies from lifestyle adjustment to heart transplantation; its aims are to reduce HF symptoms, prolong patient survival and minimize risk [2]. One alternative available in the market for HF treatment is Left Ventricular Assist Device (LVAD). Chronic Intermittent Mechanical Support (CIMS) device is a novel (LVAD) heart failure treatment using counterpulsation similar to Intra Aortic Balloon Pumps (IABP). However, the implantation site of the CIMS balloon is in the ascending aorta just distal to aortic valve contrasted with IABP in the descending aorta. Counterpulsation coupled with implantation close to the aortic valve enables comparable flow augmentation with reduced balloon volume. Two prototypes of the CIMS balloon were constructed using rapid prototyping: the straight-body model is a cylindrical tube with a silicone membrane lining with zero expansive compliance. The compliant-body model had a bulging structure that allowed the membrane to expand under native systolic pressure increasing the device’s static compliance to 1.5 mL/mmHg. This study examined the effect of device compliance and vascular compliance on counterpulsating flow augmentation. Both prototypes were tested on a two-element Windkessel model human mock circulatory loop (MCL). The devices were placed just distal to aortic valve and left coronary artery. The MCL mimicked HF with cardiac output of 3 L/min, left ventricular pressure of 85/15 mmHg, aortic pressure of 70/50 mmHg and left coronary artery flow rate of 66 mL/min. The mean arterial pressure (MAP) was calculated to be 57 mmHg. Arterial compliance was set to be1.25 mL/mmHg and 2.5 mL/mmHg. Inflation of the balloon was triggered at the dicrotic notch while deflation was at minimum aortic pressure prior to systole. Important haemodynamics parameters such as left ventricular pressure (LVP), aortic pressure (AoP), cardiac output (CO), left coronary artery flowrate (QcorMean), and dP (Peak aortic diastolic augmentation pressure – AoPmax ) were simultaneously recorded for both non-assisted mode and assisted mode. ANOVA was used to analyse the effect of both factors (balloon and arterial compliance) to flow augmentation. The results showed that for cardiac output and left coronary artery flowrate, there were significant difference between balloon and arterial compliance at p < 0.001. Cardiac output recorded maximum output at 18% for compliant body and stiff arterial compliance. Left coronary artery flowrate also recorded around 20% increase due to compliant body and stiffer arterial compliance. Resistance to blood ejection recorded highest difference for combination of straight body and stiffer arterial compliance. From these results it is clear that both balloon and arterial compliance are statistically significant factors for flow augmentation on peripheral artery and reduction of resistance. Although the result for resistance reduction was different from flow augmentation, these results serves as an important aspect which will influence the future design of the CIMS balloon and its control strategy. References: 1. Mathers C, Boerma T, Fat DM. The Global Burden of disease:2004 update. Geneva: World Heatlh Organization; 2008. 2. Jessup M, Brozena S. Heart Failure. N Engl J Med 2003;348:2007-18.

Molecular dynamics simulations of the adsorption and diffusion behavior of pure and mixed alkanes in silicalite

The adsorption and diffusion of mixed hydrocarbon components in silicalite have been studied using molecular dynamic simulation methods. We have investigated the effect of molecular loadings and temperature on the diffusional behavior of both pure and mixed alkane components. For binary mixtures with components of similar sizes, molecular diffusional behavior in the channels was noticed to be reversed as loading is increased. This behavior was noticeably absent for components of different sizes in the mixture. Methane molecules in the methane/propane mixture have the highest diffusion coefficients across the entire loading range. Binary mixtures containing ethane molecules prove more difficult to separate compared to other binary components. In the ternary mixture, however, ethane molecules diffuse much faster at 400 K in the channel with a tendency to separate out quickly from other components. © 2005 Elsevier Inc. All rights reserved.

Effects of microstructure, temperature and R-ratio on fatigue crack propagation and threshold behaviour in two Ni-base alloys

Fatigue crack propagation and threshold data for two Ni-base alloys, Astroloy and Nimonic 901, are reported. At room temperature the effect which altering the load ratio (R-ratio) has on fatigue behaviour is strongly dependent on grain size. In the coarse grained microstructures crack growth rates increase and threshold values decrease markedly as R rises from 0. 1 to 0. 8, whereas only small changes in behaviour occur in fine grained material. In Astroloy, when strength level and gamma grain size are kept constant, there is very little effect of processing route and gamma prime distribution on room temperature threshold and crack propagation results. The dominant microstructural effect on this type of fatigue behaviour is the matrix ( gamma ) grain size itself.

Measuring employee innovation:a review of existing scales and the development of the innovative behavior and innovation support inventories across cultures

Purpose - The paper develops a model of employee innovative behavior conceptualizing it as distinct from innovation outputs and as a multi-faceted behavior rather than a simple count of ‘innovative acts’ by employees. It understands individual employee innovative behaviors as a micro-foundation of firm intrapreneurship that is embedded in and influenced by contextual factors such as managerial, organizational and cultural support for innovation. Building from a review of existing employee innovative behavior scales and theoretical considerations we develop and validate the Innovative Behavior Inventory (IBI) and the Innovation Support Inventory (ISI). Design/methodology/approach – Two pilot studies, a third validation study in the Czech Republic and a fourth cross-cultural validation study using population representative samples from Switzerland, Germany, Italy and the Czech Republic (N=2812 employees and 450 entrepreneurs) were conducted. Findings - Both inventories were reliable and showed factorial, criterion, convergent and discriminant validity as well as cross-cultural equivalence. Employee innovative behavior was supported as comprising of idea generation, idea search, idea communication, implementation starting activities, involving others and overcoming obstacles. Managerial support was the most proximal contextual influence on innovative behavior and mediated the effect of organizational support and national culture. Originality/value - The paper advances our understanding of employee innovative behavior as a multi-faceted phenomenon and the contextual factors influencing it. Where past research typically focuses on convenience samples within a particular country, we offer first robust evidence that our model of employee innovative behavior generalizes across cultures and types of samples. Our model and the IBI and ISI inventories enable researchers to build a deeper understanding of the important micro-foundation underpinning intrapreneurial behavior in organizations and allow practitioners to identify their organizations’ strengths and weaknesses related to intrapreneurship.

Influence of Forecasting, Planning and Controlling on the performance of Small and Medium Enterprises

Production Planning and Control (PPC) systems have grown and changed because of the developments in planning tools and models as well as the use of computers and information systems in this area. Though so much is available in research journals, practice of PPC is lagging behind and does not use much from published research. The practices of PPC in SMEs lag behind because of many reasons, which need to be explored. This research work deals with the effect of identified variables such as forecasting, planning and control methods adopted, demographics of the key person, standardization practices followed, effect of training, learning and IT usage on firm performance. A model and framework has been developed based on literature. Empirical testing of the model has been done after collecting data using a questionnaire schedule administered among the selected respondents from Small and Medium Enterprises (SMEs) in India. Final data included 382 responses. Hypotheses linking SME performance with the use of forecasting, planning and controlling were formed and tested. Exploratory factor analysis was used for data reduction and for identifying the factor structure. High and low performing firms were classified using a Logistic Regression model. A confirmatory factor analysis was used to study the structural relationship between firm performance and dependent variables.

Assessing the influence of abiotic factors and leaf-level properties on the stability of growing-season canopy greenness in a deciduous forest

Maps depicting spatial pattern in the stability of summer greenness could advance understanding of how forest ecosystems will respond to global changes such as a longer growing season. Declining summer greenness, or “greendown”, is spectrally related to declining near-infrared reflectance and is observed in most remote sensing time series to begin shortly after peak greenness at the end of spring and extend until the beginning of leaf coloration in autumn,. Understanding spatial patterns in the strength of greendown has recently become possible with the advancement of Landsat phenology products, which show that greendown patterns vary at scales appropriate for linking these patterns to proposed environmental forcing factors. This study tested two non-mutually exclusive hypotheses for how leaf measurements and environmental factors correlate with greendown and decreasing NIR reflectance across sites. At the landscape scale, we used linear regression to test the effects of maximum greenness, elevation, slope, aspect, solar irradiance and canopy rugosity on greendown. Secondly, we used leaf chemical traits and reflectance observations to test the effect of nitrogen availability and intrinsic water use efficiency on leaf-level greendown, and landscape-level greendown measured from Landsat. The study was conducted using Quercus alba canopies across 21 sites of an eastern deciduous forest in North America between June and August 2014. Our linear model explained greendown variance with an R2=0.47 with maximum greenness as the greatest model effect. Subsequent models excluding one model effect revealed elevation and aspect were the two topographic factors that explained the greatest amount of greendown variance. Regression results also demonstrated important interactions between all three variables, with the greatest interaction showing that aspect had greater influence on greendown at sites with steeper slopes. Leaf-level reflectance was correlated with foliar δ13C (proxy for intrinsic water use efficiency), but foliar δ13C did not translate into correlations with landscape-level variation in greendown from Landsat. Therefore, we conclude that Landsat greendown is primarily indicative of landscape position, with a small effect of canopy structure, and no measureable effect of leaf reflectance. With this understanding of Landsat greendown we can better explain the effects of landscape factors on vegetation reflectance and perhaps on phenology, which would be very useful for studying phenology in the context of global climate change

Study of antioxidant, antiproliferative and apoptosis-inducing properties of wild mushrooms from the Northeast of Portugal

Mushrooms are known as a powerful source of bioactive compounds including antioxidants, inhibitors of human tumour cell lines growth, inducers of apoptosis and enhancers of immunity. Indeed, many pre-clinical studies have been conducted in human tumour cell lines and in some cases a number of compounds isolated from mushrooms have followed to clinical trials. The Northeast of Portugal is one of the European regions with higher wild mushrooms diversity. However, to our knowledge, no studies had been conducted so far to verify their bioactivities. The main aim of this work was the evaluation of the bioactive properties (antioxidant properties and growth inhibitory potential on human tumour cell lines) of wild edible mushrooms collected in the Northeast of Portugal. Once properly identified, methanolic, ethanolic and boiling water extracts were prepared from thirty eight wild mushroom species collected in that region. Chemical characterization was obtained by high performance liquid chromatography (HPLC) coupled to a photodiode array detector (DAD) or to a refraction index detector (RI). Antioxidant activity assays were carried out in those extracts, including evaluation of 2,2-diphenyl-1-picrylhydrazyl (DPPH) radicals scavenging capacity, reducing power and inhibition of β-carotene bleaching. Extract-induced cell growth inhibition was assessed with the sulforhodamine B assay in four human tumour cell lines (NCI-H460 - lung cancer, MCF-7 -breast cancer, HCT-15 -colon cancer and AGS - gastric cancer). The effects on cell cycle profile and apoptosis were evaluated by flow cytometry and the effect on the expression levels of proteins related to cell cycle and apoptosis was further investigated by Western blotting. Three wild edible mushroom species revealed growth inhibitory activity in the studied human tumour cell lines: Clitocybe alexandri ethanolic extract, Lepista inversa methanolic extract and Suillus collinitus methanolic extract. C. alexandri ethanolic extract induced an S-phase cell cycle arrest and increased the percentage of apoptotic cells, in the NCI-H460 cell line. The analysed mushroom species also provided interesting antioxidant potential, mainly the boiling water extract of L. inversa which showed the highest DPPH radical scavenging activity, reducing power and β-carotene bleaching inhibition. S. collinitus methanolic extract induced a slight increase in the number of cells in G1, with a concomitant decrease in the percentage of cells in the S phase of the cell cycle and an increase in the percentage of apoptotic cells, in the MCF-7 cell line. The combined use of the S. collinitus methanolic extract and etoposide caused a greater decrease in the percentage of cell growth, when compared to either of them used individually, indicating the potential benefit of this combination. The tested extracts were chemically characterized and protocatechuic, p-hydroxybenzoic, p-coumaric and cinnamic acids were the main compounds identified on the phenolic (methanolic and ethanolic) extracts, while mannitol, trehalose and arabinose were the main sugars found in the polysaccharidic (boiling water) extracts after hydrolysis. The individual compounds identified in the extracts were submitted to a screening of tumour cells growth inhibitory activity, but only the phenolic acids and a related compound, cinnamic acid, presented activity. This compound was found to be the most potent one regarding cell growth inhibition in the NCI-H460 cell line. The effect of the individual and combined treatment with the identified compounds was also evaluated. Cinnamic and protochatequic acids caused a statistically significantly reduction in the number of viable cells. In addition, p-hydroxybenzoic acid did not show any significantly reduction in the viable cell number. Nevertheless, it was verified that the concomitant use of the three compounds provided the strongest decrease in the viable cell number, suggesting a possible concomitant effect of those compounds. Overall, the present work has contributed to further understand the bioactive potential of wild edible mushrooms from the Northeast of Portugal. This study allowed to identify some species with antioxidant or tumour cell growth inhibitory potential.

Differential mechanisms of angiotensin II and PDGF-BB on migration and proliferation of coronary artery smooth muscle cells

Angiotensin II (Ang II) and platelet-derived growth factor-BB (PDGF-BB) are associated with excessive cell migration, proliferation and many growth-related diseases. However, whether these agents utilise similar mechanisms to trigger vascular pathologies remains to be explored. The effects of Ang II and PDGF-BB on coronary artery smooth muscle cell (CASMC) migration and proliferation were investigated via Dunn chemotaxis assay and the measurement of [3H]thymidine incorporation rates, respectively. Both atherogens produced similar degrees of cell migration which were dramatically inhibited by mevastatin (10 nM). However, the inhibitory effects of losartan (10 nM) and MnTBAP (a free radical scavenger; 50 μM) were found to be unique to Ang II-mediated chemotaxis. In contrast, MnTBAP, apocynin (an antioxidant and phagocytic NADPH oxidase inhibitor; 500 μM), mevastatin and pravastatin (100 nM) equally suppressed both Ang II and PDGF-BB-induced cellular growth. Although atherogens produced similar changes in NADPH oxidase, NOS and superoxide dismutase activities, they differentially regulated antioxidant glutathione peroxidase activity which was diminished by Ang II and unaffected by PDGF-BB. Studies with signal transduction pathway inhibitors revealed the involvement of multiple pathways i.e. protein kinase C, tyrosine kinase and MAPK in Ang II- and/or PDGF-BB-induced aforementioned enzyme activity changes. In conclusion, Ang II and PDGF-BB may induce coronary atherosclerotic disease formation by stimulating CASMC migration and proliferation through agent-specific regulation of oxidative status and utilisation of different signal transduction pathways.

Memory-enhancing effect of Rhodiola rosea L extract on aged mice

Purpose: The memory-enhancing effects of Rhodiola rosea L. extract (RRLE) on normal aged mice were assessed. Methods: In the open-field test, the effect of RRLE (150 and 300 mg/kg) on mouse locomotive activities was evaluated by investigating the extract’s influence on CAT and AchE activities in the brain tissue of mice. Results: Compared with aged group, high dose of RRLE reduced the total distance (3212.4 ± 123.1 cm, p < 0.05) significantly, increased catalase (CAT) activity (101.4 ± 12.2 U/mg pro, p < 0.05), and inhibited acetyl cholinesterase (AChE) activity (0.94 ± 0.12 U/mg pro, p < 0.05) in the brain tissue of aged mice. Conclusion: The results show that RRLE improves the memory functions of aged mice probably by increasing CAT activity while decreasing AChE activity.

The effect of increased light intensity on the aggressive behavior of the Nile tilapia, Oreochromis niloticus (Teleostei: Cichlidae)

Animals show behavioral and physiological changes that emerge in response to environmental perturbations (i.e., emergency life-history stages). In this study, we investigate the effects of light intensity on aggressive encounters and social stability in groups of adult male Nile tilapia, Oreochromis niloticus (Linnaeus, 1758). The study compared the behavior observed under low (280.75 ± 50.60 lx) and high (1394.14 ± 520.32 lx) light intensities, with 12 replicates for each treatment. Adult fish were isolated in 36-L aquaria for 96 hours, and three males were grouped for 11 days in 140-L aquaria. Agonistic behavior was video-recorded (10 min/day) on the 3rd, 5th, 7th, and 9th day to quantify aggressive interactions and social stability. There was an effect of light intensity and day of observation on the total number of agonistic behaviors performed by the fish group. Besides, increased frequency of aggressive interactions (the sum of the four sessions) by the alpha, beta and gamma fish occurred at the higher light intensity. The dominance ranks of the fish remained unchanged across the observation sessions under both the low and high light intensities. We concluded that enhanced light intensity has a cumulative effect that increases the aggressiveness of the Nile tilapia but that this effect is not sufficiently strong to destabilize the social hierarchy.

Effect of organic amendments on the afforestation performance of plant species on degraded semiarid conditions

Vegetation plays a fundamental role in soil conservation, so it is common to consider an increase in vegetation cover as one of the techniques to mitigate the effects of desertification in Mediterranean forest environments. There are two factors limiting the establishment and growth of seedlings in dry environments: (i) an excessive radiation and, (ii) the limited availability of water during the summer drought. During an afforestation plan, soil preparation is always necessary to reduce sapling mortality. The goal of this study was to analyze the effect of various organic amendments on soil according to chemical and hydrological properties, and to assess the effects of these parameters on an afforestal proposal under Mediterranean climate conditions. Five amendments were applied in an experimental set of plots: straw mulching (SM); mulch with chipped branches of Aleppo Pine (PM); TerraCotten hydroabsobent polymers (HP); sewage sludge (RU); sheep manure (SH) and control (C). Plots were afforested following the same spatial pattern, and amendments were mixed with the soil at the rate 10 Mg ha -1 . Under bare soil conditions (C), most of mortalities occurred during the summer period of the first year. A substantial positive effect of SM, PM and HP on the survival rates have been clearly observed. Conversely, when the soil was amended with SH, the survival rate quickly decreased or remained more or less constant regarding to C. In this study, the lack of differences on chemical properties indicates that there may exist other reasons to justify the differences that were found in the pattern of vegetation. However, regarding to the hydrological properties some differences have been found. In C, soils were registered below the wilting point during 4 months a year, and therefore, in the area of water unusable by plants. These months were coinciding with the summer Mediterranean drought and can justify the high mortality found on plants. Conversely, in SM, PM and HP, soil moisture remained below the wilting point less period than C and, the plant available water was also higher. In these treatments, the survival sapling rates measured were the highest. SH showed water holding capacity slightly more limited than C. For this treatment, the survival sapling rates measured were the lowest. In conclusion, from a land management standpoint, the PM, SM and HP have been proved as a significant method to reduce sapling mortality rates during the Mediterranean summer drought.

Predictions of J integral and tensile strength of clay/epoxy nanocomposites material using phase field model

We predict macroscopic fracture related material parameters of fully exfoliated clay/epoxy nano- composites based on their fine scale features. Fracture is modeled by a phase field approach which is implemented as user subroutines UEL and UMAT in the commercial finite element software Abaqus. The phase field model replaces the sharp discontinuities with a scalar damage field representing the diffuse crack topology through controlling the amount of diffusion by a regularization parameter. Two different constitutive models for the matrix and the clay platelets are used; the nonlinear coupled system con- sisting of the equilibrium equation and a diffusion-type equation governing the phase field evolution are solved via a NewtoneRaphson approach. In order to predict the tensile strength and fracture toughness of the clay/epoxy composites we evaluated the J integral for different specimens with varying cracks. The effect of different geometry and material parameters, such as the clay weight ratio (wt.%) and the aspect ratio of clay platelets are studied.