<p>This thesis consists of three papers that examine college choice and earnings among university graduates in Sweden.</p><p>Paper [I] analyzes how geographical accessibility to higher education affects university enrollment decisions in Sweden. The empirical findings show that the probability of enrollment in university education increases with accessibility to university education. The results also indicate that accessibility adds to the likelihood of attending a university within the region of residence. Both these findings are robust with regard to different specifications of accessibility. The empirical results furthermore indicate that the enrollment decisions of individuals with a less privileged background are more sensitive to accessibility to university education than are the decisions of individuals from a more favorable background.</p><p>Paper [II] examines the effect on earnings of graduating from five different college groups. The paper relies on selection on observables and linear regression to identify the earnings effect of college choice. Contrary to the majority of previous Swedish studies, we do not find any systematic differences in estimated earnings between college graduates from the different college groups. This finding does not only hold when considering all college graduates, but also when focusing on men and women separately as well as when considering college graduates in two specific fields of education. The results suggest that an estimator of the earnings effects of college choice that does not properly adjust for ability is likely to be substantially biased.</p><p>Paper [III] estimates the causal effect on earnings of graduating from old universities rather than new universities/university colleges. The study compares estimates from several different matching methods and linear regression. We cannot find any significant differences in earnings between graduates from the two groups of colleges. This holds for male and female sub-samples covering all majors, as well as male and female sub-samples covering two broad fields of education. The results are robust with regard to different methods of propensity score matching and regression adjustment. Furthermore, the results indicate little sensitivity with regard to the empirical support in the data and alternative specifications of the propensity scores.</p>

<p>In this paper I study how the repeal of the Swedish wealth tax (1 of January 2007) has affected people´s labour supply behaviour. This particular issue is relevant because it may help us understand some of the effects of the earnings tax changes that have taken place in Sweden. Accoring to standard economic theory a repealed wealth tax is similar to an income effect for the persons who previously paid the tax. That means that they theoretically will want to consume more leisure, that is decrease their labour supply. The method I am using to test this hypothesis is a difference-in-difference approach where the treatment group consists of persons who previously paid the tax and the control group of comparable persons who did not pay the tax. The data I am using is taken from a Swedish database called LINDA, compiled by the Swedish Central Agency for Statistics (SCB). My main result in this paper is that the repealed wealth tax does not seem to have had any influnece on the labour supply behavior of the persons who previously paid the tax.</p>

<p>Using data from the Labour Force Survey conducted by Statistics South Africa twice yearly this thesis intend to evaluate the Expanded Public Works Programme regarding its effectiveness in creating employment and raising income in households with participating individuals. The South African labour market is well known for its high rates of unemployment and also its segregation, primarily between black and white people, but also young people are having a hard time finding jobs. In order to fight these problems the South African government has launched the Expanded Public Works Programme(EPWP) which provides low- semi-skilled labour with short term employment, the primary target groups being black and coloured people, women, disabled people and young people. Our findings indicate that the EPWP does not significantly enhance the individual’s probability of being employed, nor does it raise the per capita income of households with participating individuals.</p>

This dissertation addresses a number of inter-related and fundamental problems in computer vision. Specifically, we address object discovery, recognition, segmentation, and 3D pose estimation in images, as well as 3D scene reconstruction and scene interpretation. The key ideas behind our approaches include using shape as a basic object feature, and using structured prediction modeling paradigms for representing objects and scenes. In this work, we make a number of new contributions both in computer vision and machine learning. We address the vision problems of shape matching, shape-based mining of objects in arbitrary image collections, context-aware object recognition, monocular estimation of 3D object poses, and monocular 3D scene reconstruction using shape from texture. Our work on shape-based object discovery is the first to show that meaningful objects can be extracted from a collection of arbitrary images, without any human supervision, by shape matching. We also show that a spatial repetition of objects in images (e.g., windows on a building facade, or cars lined up along a street) can be used for 3D scene reconstruction from a single image. The aforementioned topics have never been addressed in the literature. The dissertation also presents new algorithms and object representations for the aforementioned vision problems. We fuse two traditionally different modeling paradigms Conditional Random Fields (CRF) and Random Forests (RF) into a unified framework, referred to as (RF)^2. We also derive theoretical error bounds of estimating distribution ratios by a two-class RF, which is then used to derive the theoretical performance bounds of a two-class (RF)^2. Thorough experimental evaluation of individual aspects of all our approaches is presented. In general, the experiments demonstrate that we outperform the state of the art on the benchmark datasets, without increasing complexity and supervision in training. / Graduation date: 2011 / Access restricted to the OSU Community at author's request from May 12, 2011 - May 12, 2012

Graduation date: 2004

Biologists need tools to see the structural relationships encoded in biological sequences (strings). The Walking Tree heuristics calculate some of these relationships. I have designed and implemented graphic presentations which allow the biologist (user) to see these relations. This thesis contains background information on the biological sequences and some background on the Walking Tree heuristics. I demonstrate my methods by showing a visual matching of mitochondrial genomes. I also show matchings based on amino acids and on hydrophobicity. I also show how the parameters of the visualization can be varied to produce more useful pictures. I implemented a parallel version of the Walking Tree heuristic and used it to produce a phylogenetic tree for picornaviruses. I also implemented several user interfaces. These programs are available on my WWW page which allows a user to produce a picture of a matching by giving the sequences in Gen Bank format and by making a few mouse clicks. / Graduation date: 1998

This paper describes a method by which range data from a sonar or other type of rangefinder can be used to determine the 2-dimensional position and orientation of a mobile robot inside a room. The plan of the room is modeled as a list of segments indicating the positions of walls. The method works by extracting straight segments from the range data and examining all hypotheses about pairings between the segments and walls in the model of the room. Inconsistent pairings are discarded efficiently by using local constraints based on distances between walls, angles between walls, and ranges between walls along their normal vectors. These constraints are used to obtain a small set of possible positions, which is further pruned using a test for physical consistency. The approach is extremely tolerant of noise and clutter. Transient objects such as furniture and people need not be included in the room model, and very noisy, low-resolution sensors can be used. The algorithm's performance is demonstrated using Polaroid Ultrasonic Rangefinder, which is a low-resolution, high-noise sensor.

We present an approach for image database retrieval using a very large number of highly-selective features and simple on-line learning. Our approach is predicated on the assumption that each image is generated by a sparse set of visual "causes" and that images which are visually similar share causes. We propose a mechanism for generating a large number of complex features which capture some aspects of this causal structure. Boosting is used to learn simple and efficient classifiers in this complex feature space. Finally we will describe a practical implementation of our retrieval system on a database of 3000 images.

Present-day economic and environmental constraints push power systems to be operated closer to their limits. A common limiting factor for power transmission is the risk of voltage instability in recent years. As the ultimate countermeasure to voltage collapse, load shedding is normally considered the last resort, when there are no other alternatives to stop an approaching voltage collapse. The requirements of a practical load shedding scheme are to prevent a power system from voltage collapse and to maximize its reliability. In order to design such a scheme, the following tasks are equally important: 1. Recognizing the approaching voltage collapse. 2. Determining the best load shedding locations. 3. Minimizing the amount of load shedding. This thesis firstly investigates the widely used undervoltage load shedding schemes (UVLS) and the single-port impedance match (SPIM) based schemes. The findings explain the difficulties faced by them. An original load shedding oriented voltage stability monitoring scheme, which involves developing a new multi-port network equivalent, is then developed. With the help of the multi-port network equivalent, the monitoring scheme can not only recognize the approaching voltage collapse in time, but also can easily rank the load buses based on their weakness. The results of ranking are consistent with those obtained from modal analysis method. This thesis then proposes a practical event-driven load shedding scheme based on the experiences learned from the schemes implemented by various utilities. The scheme involves developing a multistage method, which is to optimize the amount of load shedding. A general design procedure for the scheme is presented in the thesis. Using a real 2038 bus system as an example, the design methodology is described in detail. The methodology is expected to help power system engineers develop their own load shedding schemes. A practical emergency demand response scheme is also developed and presented in the appendix. It is aimed at choosing the proper demand response participants and minimizing the total cost while achieving a certain level of operation reserves. / Power Engineering and Power Electronics

This research presents a procedure for interactive segmentation and automatic tracking of moving objects in a video sequence. The user outlines the region of interest (ROI) in the initial frame; the procedure builds a refined mask of the dominant object within the ROI. The refined mask is used to model a spline template of the object to be tracked. The tracking algorithm then employs a motion model to track the template through a sequence of frames and gathers the 3D affine motion parameters of the object from each frame. The extracted template is compared with a previously stored library of 3D shapes to determine the closest 3D object. If the extracted template is completely new, it is used to model a new 3D object which is added to the library. To recreate the motion, the motion parameters are applied to the 3D object in a virtual environment. The procedure described here can be applied to industrial problems such as traffic management and material flow congestion analysis.