Browsing by Author "Shcherbakov, Dmitry"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Application of reinforcement learning to autonomous aircraft in partially observable environments
    (University of New Brunswick, 2018) Shcherbakov, Dmitry; Fleming, Michael
    This thesis provides a brief survey of the mathematical background of the reinforcement learning (RL) method and sketches the current state of arguably the most developed area of RL application, to the problem of controlling autonomous vehicles (self-driven car-like vehicles). This is then compared to RL solutions in autonomous piloting tasks. Contrasting the two shows that the latter may benefit from a common framework for RL applications. We propose a framework for autonomous piloting tasks, provide a detailed description of the toolkit available for the framework, and perform an experiment with described instruments. The experiment is designed to determine whether a small fixed memory window can mitigate the adverse influence of such unobserved factors as wind bursts and turbulence. Tests show that the memory mechanism that encapsulates control feedback is an informative input for the learning agent, as long as the unobserved factors affect control behavior significantly.
  • Thumbnail Image
    Item
    Decomposing the gender wage gap
    (University of New Brunswick, 2017) Shcherbakov, Dmitry; McDonald, Ted
    The gender wage gap persists as an important and controversial topic in various socio-economic agendas and appears as a topic of research in scholarly research regularly. The conventional approach to conducting a gender wage gap assessment is measuring the difference in average earnings between men and women that remains after controlling for various explanatory variables suggested by theory and empirical labor studies. The list of explanatory factors used in particular papers varies depending on the specificity of the population sample, availability of data, research focus and other aspects. Specifically, variation in occupational distribution between men and women in most cases in the literature has been limited to 20-40 fairly broad occupational groups. The inclusion of occupation as an explanatory variable is a controversial approach as it may lead to an undervaluation of the influence of labor market gender segregation tendencies, or, on the opposite, to over-justification of gender-pay differentials. This thesis provides an overview of conventional techniques used for gender wage gap estimations, discusses the importance and appropriateness of occupation as an explanatory variable, and contrasts gender wage gap estimations with and without these controls in the context of the Canadian labor market. Furthermore the thesis tests whether very detailed occupation controls is advantageous compared with a more limited set of broad occupational groups in terms of examination of the wage gap, and suggests evidence for justification of occupational gender segregation that is often perceived as discrimination.