University of Mannheim

About: University of Mannheim is a education organization based out in Mannheim, Germany. It is known for research contribution in the topics: Context (language use) & Politics. The organization has 4448 authors who have published 12918 publications receiving 446557 citations. The organization is also known as: Uni Mannheim & UMA.

Probabilistic subgraph matching based on convex relaxation

Abstract: We present a novel approach to the matching of subgraphs for object recognition in computer vision. Feature similarities between object model and scene graph are complemented with a regularization term that measures differences of the relational structure. For the resulting quadratic integer program, a mathematically tight relaxation is derived by exploiting the degrees of freedom of the embedding space of positive semidefinite matrices. We show that the global minimum of the relaxed convex problem can be interpreted as probability distribution over the original space of matching matrices, providing a basis for efficiently sampling all close-to-optimal combinatorial matchings within the original solution space. As a result, the approach can even handle completely ambiguous situations, despite uniqueness of the relaxed convex problem. Exhaustive numerical experiments demonstrate the promising performance of the approach which – up to a single inevitable regularization parameter that weights feature similarity against structural similarity – is free of any further tuning parameters.

Belief elicitation in experiments: is there a hedging problem?

Abstract: Belief-elicitation experiments usually reward accuracy of stated beliefs in addition to payments for other decisions. But this allows risk-averse subjects to hedge with their stated beliefs against adverse outcomes of the other decisions. So can we trust the existing belief-elicitation results? And can we avoid potential hedging confounds? We propose an experimental design that theoretically eliminates hedging opportunities. Using this design, we test for the empirical relevance of hedging effects in the lab. Our results suggest that hedging confounds are not a major problem unless hedging opportunities are very prominent. If hedging opportunities are transparent, and incentives to hedge are strong, many subjects do spot hedging opportunities and respond to them. The bias can go beyond players actually hedging themselves, because some expect others to hedge and best respond to this.

Numerical Solution of Fractional Order Differential Equations by Extrapolation

Abstract: We present an extrapolation type algorithm for the numerical solution of fractional order differential equations. It is based on the new result that the sequence of approximate solutions of these equations, computed by means of a recently published algorithm by Diethelm [6], possesses an asymptotic expansion with respect to the stepsize. From this we conclude that the application of extrapolation is justified, and we obtain a very efficient differential equation solver with practically no additional numerical costs. This is also illustrated by a number of numerical examples.

When Less is More Selection Problems in Large- N and Small- N Cross-National Comparisons

Abstract: The problem of case selection is a crucial but often overlooked issue in comparative cross-national research. The article discusses methodological shortcomings and potential solutions in selecting cases. All comparative research of social entities, whether quantitative or qualitative, faces the problem of contingency, the fact that the potential pool of cases has been pre-selected by historical and political processes. In large-N cross-national studies the use of inference statistics is problematic since random selection is rarely given and the cases represent a highly stratified set. In small-N case studies, however, the selection of cases is a deliberate choice based on the theory-driven comparative method. The epistemological and methodological problems of both comparative approaches are discussed and evaluated.

Credit Line Usage, Checking Account Activity, and Default Risk of Bank Borrowers

Abstract: We investigate the link between account activity and information production on borrower quality. Based on a unique data set, we find that credit line usage, limit violations, and cash inflows exhibit abnormal patterns approximately 12 months before default events. Measures of account activity substantially improve default predictions and are especially helpful for monitoring small businesses and individuals. We also find that the early warning indications from account activity result in higher loan spreads, and in a higher likelihood of limit reductions and complete write-offs. Our results highlight that the information on account activity provides banks with a real-time window into the borrower’s cash flows, creating a unique advantage over non-bank lenders and capital markets.