Refine
Document Type
- Article (1)
- Bachelor Thesis (1)
- Master's Thesis (1)
Language
- English (3)
Has Fulltext
- yes (3)
Keywords
- Bayesian neural network (1)
- Botswana (1)
- Deep learning (1)
- MSP Botswana (1)
- Maun Science Park (1)
- Transformation models (1)
- Variational inference (1)
Institute
Botswana, a new construction project – the Maun Science Park - is to be built with a focus on sustainability and to create a new living space for the rapidly growing population in Africa. The project will be a blueprint for future projects in Africain terms of progress, technology and sustainability. This thesis will deal with its financial framework and will serve as a basis for the development of ways and means of financing such projects.
Botswana is a country in southern Africa with rich mineral resources, which has built its economy on mining. Due to challenges in the upcoming years caused by climate and demographic change, it aims to move away from a resource-based economy to a knowledge-based economy in the long term. In order to support the
process, the Maun Science Park, a centre for research and development is planned to be created in Maun, a town on the edge of the Okavango Delta. The project is initiated by the “International Resilience and Sustainability Partnership” (inRES), a non-governmental organization. The project is currently in the initiation phase.
The purpose of this thesis is to determine a cost framework with exemplary developer calculation and sensitivity analysis for the Maun Science Park Project in Botswana. Therefor, a source research was performed in a first step. Based on this, interviews were conducted with members of the inRES. Based on the data
obtained and further assumptions, a cost framework for the different project phases of the MSP project was established. Subsequently, a developer calculation
was exemplarily carried out on the basis of the project phase 2 and a sensitivity analysis was performed.
During the interviews, data was collected on the different project phases. It became clear that the interview partners had partly inconsistent perceptions
about different project phases. The calculation can be used as a basis for further calculation at the time of concretization of the planning data.
Black-box variational inference (BBVI) is a technique to approximate the posterior of Bayesian models by optimization. Similar to MCMC, the user only needs to specify the model; then, the inference procedure is done automatically. In contrast to MCMC, BBVI scales to many observations, is faster for some applications, and can take advantage of highly optimized deep learning frameworks since it can be formulated as a minimization task. In the case of complex posteriors, however, other state-of-the-art BBVI approaches often yield unsatisfactory posterior approximations. This paper presents Bernstein flow variational inference (BF-VI), a robust and easy-to-use method flexible enough to approximate complex multivariate posteriors. BF-VI combines ideas from normalizing flows and Bernstein polynomial-based transformation models. In benchmark experiments, we compare BF-VI solutions with exact posteriors, MCMC solutions, and state-of-the-art BBVI methods, including normalizing flow-based BBVI. We show for low-dimensional models that BF-VI accurately approximates the true posterior; in higher-dimensional models, BF-VI compares favorably against other BBVI methods. Further, using BF-VI, we develop a Bayesian model for the semi-structured melanoma challenge data, combining a CNN model part for image data with an interpretable model part for tabular data, and demonstrate, for the first time, the use of BBVI in semi-structured models.
