Em diferentes aplicações de Aprendizado de Máquina podemos estar interessados na minimização do valor esperado de certa função de perda. Para a resolução desse problema, Otimização estocástica e Sample Size Selection têm um papel importante. No presente trabalho se apresentam as análises teóricas de alguns algoritmos destas duas áreas, incluindo algumas variações que consideram redução da variância. Nos exemplos práticos pode-se observar a vantagem do método Stochastic Gradient Descent em relação ao tempo de processamento e memória, mas, considerando precisão da solução obtida juntamente com o custo de minimização, as metodologias de redução da variância obtêm as melhores soluções. Os algoritmos Dynamic Sample Size Gradient e Line Search with variable sample size selection apesar de obter soluções melhores que as de Stochastic Gradient Descent, a desvantagem se encontra no alto custo computacional deles. / In different Machine Learnings applications we can be interest in the minimization of the expected value of some loss function. For the resolution of this problem, Stochastic optimization and Sample size selection has an important role. In the present work, it is shown the theoretical analysis of some algorithms of these two areas, including some variations that considers variance reduction. In the practical examples we can observe the advantage of Stochastic Gradient Descent in relation to the processing time and memory, but considering accuracy of the solution obtained and the cost of minimization, the methodologies of variance reduction has the best solutions. In the algorithms Dynamic Sample Size Gradient and Line Search with variable sample size selection, despite of obtaining better solutions than Stochastic Gradient Descent, the disadvantage lies in their high computational cost.

Post-copulatory sexual selection is a strong evolutionary force, affecting morphological and behavioural traits in males and females in species with polyandrous mating systems. Many insects are subject to sperm competition; sperm from rival males compete to fertilise ova. Since sperm are finite, males should allocate them economically, tailoring ejaculate allocation to suit the reproductive potential of individual matings. Theory suggests when sperm competition risk is high, males should increase sperm numbers to achieve greater reproductive success than their rivals, but evidence of this expected fitness consequence of ejaculate allocation is largely lacking. In this thesis, I use Callosobruchus maculatus beetles to investigate the causes of ejaculate allocation patterns, and to examine whether ejaculate allocation does affect male reproductive success. In Chapter 3, I investigate the effect of rival male presence on ejaculate size and find that, while males grouped with rivals as adults produce bigger ejaculates, their increased effort unexpectedly does not lead to increased reproductive success. In Chapter 4, I examine whether larval conditions also affect ejaculate size, and find that, contrary to sperm competition theory, males reared under dense conditions produce smaller ejaculates than those reared solitarily, and that male reproductive success is consequently elevated in males reared at low larval densities compared to those reared at high densities. In Chapter 5, I then demonstrate that ejaculates produced by low density males contain more sperm than ejaculates produced by high density males, suggesting males do not respond to sperm competition level represented by larval density, but instead suffer resource limitation when reared at high density. In Chapter 6, I investigate the effects of water provision on ejaculate size, and find that males given water produce larger ejaculates, and females given water receive smaller ejaculates. Finally, I link my findings with those of other studies, and suggest my most important result is that plasticity of ejaculate allocation cannot be assumed to be an adaptive behaviour; studies directly measuring the fitness effects of male ejaculate allocation are needed, even when observed patterns conform to theory.

As commercially manufactured nanomaterials become more commonplace, they have the potential to enter ecological and biological environments sometime during their lifecycle of production, distribution, use or disposal. Despite rapid advances in the production and application of nanomaterials, little is known about how nanomaterials may interact with the environment or affect human health. This research investigates an environmental application of nanomaterials and characterizes the physicochemical properties of commonly manufactured nanomaterials in environmental health and safety studies. Characterization of nanomaterials for applications and environmental health and safety studies is essential in order to understand how physicochemical properties correlate with chemical, ecological, or biological response or lack of response. Full characterization includes determining the bulk and surface properties of nanomaterials. Bulk characterization methods examine the shape, size, phase, electronic structure and crystallinity, and surface characterization methods include surface area, arrangement of surface atoms, surface electronic structure, surface composition and functionality. This work investigates the selective catalytic reduction (SCR) of NO2 to N2 and O2 with ammonia on nanocrystalline NaY, Aldrich NaY and nanocrystalline CuY using in situ Fourier transform infrared (FTIR) spectroscopy. It was determined that the kinetics of SCR were 30% faster on nanocrystalline NaY compared to commercial NaY due to an increase in external surface area and external surface reactivity. Copper-cation exchanged nanocrystalline Y resulted in an additional increase in the rate of SCR as well as distinct NO2 and NH3 adsorption sites associated with the copper cation. These superior materials for reducing NOx could contribute to a cleaner environment. This work consists of characterization of commonly manufactured or synthesized nanomaterials and studies of nanomaterials in specific environmental conditions. Bulk and surface characterization techniques were used to examine carbon nanotubes, titanium dioxide nanoparticles, bare silver nanoparticles and polymer-coated silver nanoparticles, and copper nanoparticles. Lithium titanate nanomaterial was collected from a manufacturing facility was also characterized to identify occupational health risks. Particle size distribution measurements and chemical composition data showed the lithium titanate nanomaterial forms larger micrometer agglomerates, while the nanoparticles present were due to incidental processes. A unique approach was applied to study particle size during dissolution of nanoparticles in aqueous and acidic conditions. An electrospray coupled to a scanning mobility particle sizer (ES-SMPS) was used to determine the particle size distribution (PSD) of bare silver nanoparticles in nitric acid and copper nanoparticles in hydrochloric acid. The results show unique, size-dependent dissolution behavior for the nanoparticles relative to their micrometer sized counterparts. This research shows size-dependent properties of nanomaterials can influence how they will be transported and transformed in specific environments, and the behavior of larger sized materials cannot be used to predict nanomaterial behavior. The type of nanomaterial and the media it enters are important factors for determining the fate of the nanomaterial. These studies will be important when considering measures for exposure control and environmental remediation of nanomaterials.

This research investigated the relationship between group size and the process of trust, cohesion, and commitment formation in joint-task groups. Specifically, the theory proposes that groups with greater six members will produce lower trust, cohesion, and commitment than those smaller than six members. Theory was developed linking group size to these processes through the mechanism of anticipated mutual perception, the amount an individual considers what others are thinking about them. Two experimental studies test the impact of anticipated mutual perception on interpersonal influence and trust, cohesion, and commitment. Study 1 tested the impact of being able to see a partner and closeness to a partner on a partner's level of influence over participants. Participants were more influenced by partners they could see than by partner's who were separated from the participant by place or time. Status differences also affected influence, even when partners were absent, having left information for use by participants. Study 2 investigated how group size affected participant reports of interpersonal trust, cohesion, and commitment to other group members. Results suggest that members of larger groups had lower commitment to other group members, and that groups larger than six members were negatively associated with levels of trust reported by group members. Further, measures used to isolate aspects of anticipated mutual perception were shown to affect the development of interpersonal trust, cohesion, and commitment, in support of the theory. Group cohesion was found to vary with expectations of competence. These expectations significantly predicted measures of shared awareness capturing aspects of anticipated mutual perception, possibly offsetting negative effects of increasing group size. Study 2 also found evidence that group members prefer to work with groups made up of 4-6 members, consistent with the theory. This research has implications for the ways in which organizations structure group work. This research suggests that the size of the group as well as the characteristics of individuals and their interaction partners may affect how people form cohesive relationships. Trust is an important aspect of groups that promotes cooperation, commitment, and collective identity formation. Further, this research supports a growing literature on the ways that micro-interaction promotes commitment to organizations by promoting or detracting from attributions of positive sentiment to smaller, nested groups within those organizations.

It has been well established that under certain circumstances wildfire is capable of producing water repellent or hydrophobic soils. Hydrophobic soils can dramatically alter runoff and erosion processes and as such have been the subject of considerable research activity. Wildfires in chaparral vegetation are recognized as being particularly susceptible to hydrophobic soil development. A comparison of chaparral fire soil heat profiles from DeBano (1989) and Weirich (unpublished) indicates that under higher fire intensity situations in chaparral a different soil heating mechanism other than just conduction heating may be at work. In contrast to the slow moving low temperature increases expected in conduction heating a much faster heat pulse resulting in more rapid temperature rises and higher temperatures at depth can also occur in chaparral wildland fires. This suggests that a better understanding of the heat transfer processes that occur at extreme fire intensities is both important and is needed. The specific aim of this study was to observe heat flow under a variety of particle sizes using a laboratory based wildfire simulator operating at intensities and durations similar to those experienced in chaparral wildfires. The wildfire simulator system consisted of a propane burner array, an array of thermocouples to measure temperatures at varying locations and depths, and a data logging system to record the results of the heating experiments. Using the simulator homogenous sand, silt, clay, and heterogeneous clay loam were subjected to 600ºC, 900ºC, and 1200ºC peak intensities with two different heating durations or treatments (H1 and H2). The heating levels and durations used were based on data from field based chaparral fire experimental temperature data previously collected by Weirich (unpublished). The system design allowed the user to control the intensity and duration of the heat treatments and the thermocouple sensor arrays measured temperatures at the flame to a soil depth of 15cm. The apparatus and experimental treatments allowed for the investigation of peak heat intensity, heat duration, slope, and most importantly particle size on heat transfer processes. The higher soil temperatures at depth, shorter times to peak temperatures at depth, and observed temperature spiking seen during some of the simulator experimental runs (specifically with respect to larger particle sizes such as sand) call into question the view that slow moving conduction may not be the only soil heat transfer process at work in high fire intensity situations such as those seen in chaparral wildfires and in particular chaparral wildfire underlain by larger particle sizes fractions such as sand.

Animals performing on simple fixed-ratio (FR) schedules vi typically pause after reinforcement delivery. The present study demonstrated systematic control of pause length in multiple fixed ratio schedules by manipulating FR size and reinforcement magnitude. In Experiment I, two adult male rabbits were stabilized on an alternating, two component, multiple FR 10 FR 10 schedule. Different colored lights were correlated with both FR components. Stability was determined and mean pause lengths were calculated as one FR component was increased in size until the schedule was FR 10 FR 50. Pausing was longer before the large FR component and was positively correlated with increases in FR size. Schedule and stimulus control were demonstrated by randomizing the order of FR presentations and by reversing the colored lights previously correlated with either the small or large FR components. Existing pause differentials were not disrupted under the light reversal or random conditions. In Experiment II, the same rabbits were stabilized at mult FR 10 FR 30 with 1/2 cc of water delivered after each component. In subsequent conditions the magnitude of reinforcement (cc's of water) delivered at the completion of the large FR component, was systematically shifted from 4 cc, to 3 cc, 2 cc, 1 cc and finally back to 1/2 cc. Pausing before the large FR was found to be inversely related to the magnitude of reinforcement delivered in that component. Clearly the animals in this experiment discriminated upcoming schedule conditions and paused relative to FR size and reinforcement magnitude. Hence, it is proper to point out that the term "post-reinforcement pause" is a misnomer since it incorrectly implies a functional relation between pause length and prior schedule conditions.

An algorithm is presented which partitions space among mapped plants according to their relative sizes and positions using one of eight rules for locating boundaries between individuals. The performance of those rules is examined using several natural and artificial data sets with diverse measures of individual size. The relative performance of the rules was the same for all natural data sets examined. The best rule, as measured by a high correlation between individual size and assigned space, placed the boundary at a distance between neighbors proportional to the relative sizes of neighbors as long as a maximum distance (also a function of size) was not exceeded. It is inferred that the algorithm identifies contact neighbors and quantifies the extent of their contact. Afield experiment is proposed to test this inference.

The results presented in this paper are from data accumulated by a study of the F3 generation of a cross between the two varieties, Sevier and New Zealand wheat. This cross was made during the summer of 1920 by Professor George Stewart with the object of combining the high-yielding power of Sevier with the strong straw of New Zealand. In the fall of 1922 the problem was assigned to me, at which time I selected about 150 superior plants, possessing the desired characteristics, at least so far as appearance was concerned. Since then these plants and the F3 generation have been studied mainly with the object of finding out whether size characters behave in Mendelian fashion or not. Other characters received attention incidentally, and will also be reported briefly. To study a variable problem with any degree of accuracy there must of necessity be a definite means by which variations may be measured. Size differences in straw are extremely difficult of measurement, and measurements at best can only be approximations. It was decided, therefore, to make a careful study of size variations in wheat heads, and only estimate straw strength. The study of this problem is valuable for two main reasons: 1. If high-yielding power and straw strength can be combined from two different varieties it is worth knowing because of the economic value of such a combination. From this point of view the original object of the cross under discussion itself justifies the investigation. 2. Since it is not a definitely settled question as to whether heritable size fluctuations behave according to the principles of Mendelism the problem has both interest and value from the point of view of genetics.

No description available.

Doctor of Philosophy (Medicine) / This work investigates a number of issues. Firstly it examines ultrasonic fetal biometry, the parameters and techniques for accurate measuring and reviews the procedure adopted for graph formation and application of regression analysis for a mathematical model to describe the relationship between fetal size and weeks of gestation. Next it establishes new Australian fetal measurement charts for the crown rump length, head circumference and abdominal circumference, based on an Australian population, to replace the charts currently in use that are over 20 years old and relate to middle class white American and British women. The new graphs, along with previous work completed by the author in 1999 on the BPD, OFD femur and humerus length, were subsequently accepted by the Australasian Society for Ultrasound in Medicine (ASUM) in 2001 as the new Australian standard for ultrasonic fetal measurements. The accuracy of first trimester ultrasound dating is also investigated, displaying the variations seen in the crown-rump length due to fetal flexion and the implications of inaccurate measuring. The third study examines inter- and intra-sonographer ultrasonic fetal measurement reproducibility in the final 6 weeks of pregnancy. The study highlights the importance of sonographer competence, standardised measuring protocols, image planes and reference charts, particularly for patients undergoing ultrasound examinations for fetal growth assessment at different practices. The fourth study looks at the incidence of fetal macrosomia and birth complications in Chinese women and Caucasian women in two time periods, 1992 and 1999/2000. The results showed a rise in macrosomic babies born to Chinese immigrants from 4% of total Chinese births in 1992 to 9.8% in 1999/2000. There was also a rise in the rate of macrosomia among Caucasian women with respective rates of 11 and 14% for the same periods. The incidence of post partum haemorrhage increased significantly over this time in both Chinese immigrant and Caucasian women. Interventions declined in all Caucasian birth-weight ranges whilst interventions for Chinese births remained stable except between 3500grams and 4000grams, where interventions rose from 35.7% to 60.5%. Fetal macrosomia is a complication of pregnancy that is increasing in incidence. One of the causes of fetal overgrowth is uncontrolled gestational diabetes mellitus and so if women thus diagnosed are closely monitored, the risks of a macrosomic baby and associated birth complications may be reduced. The final study examines the effect of gestational diabetes mellitus (GDM) on fetal growth. GDM is a complication of mid to late pregnancy caused by glucose intolerance. In the Australian population up to 8% of all pregnancies can be affected. In the Australian Chinese community the GDM rate is as high as 15% compared with 4% in Caucasian women. The risks to the fetus as a result of GDM include increased perinatal mortality, large for gestational dates, macrosomia and prematurity. The aim of this study was to determine whether the fetuses of women diagnosed with GDM were significantly larger for dates for any of the commonly ultrasonically measured fetal parameters, than in the general pregnant population. The results show that if the glycaemic levels are properly controlled, fetal size should not be compromised. The abdominal circumference measurement appears to be the important marker for fetal macrosomia, particularly in the Chinese population. The study also assessed fetal weight gain from 36 weeks gestation to term in Caucasian women with GDM and Chinese pregnancies both with and without GDM. No statistically significant difference was seen in daily weight gain between the groups investigated.