In this paper we show that complex (scale-free) network topologies naturally emerge from hyperbolic metric spaces. The hyperbolic geometry can be used to facilitate maximally efï¬...
Fragkiskos Papadopoulos, Dmitri V. Krioukov, Mari&...
—Adapting the hyperparameters of support vector machines (SVMs) is a challenging model selection problem, especially when flexible kernels are to be adapted and data are scarce....
We present a method for segmenting the parts of multiple instances of a known object category exhibiting large variations in projected shape and colour. The method builds on an ex...
Decomposition of cells in clumps is a difficult segmentation task requiring region splitting techniques. Techniques that do not employ prior shape constraints usually fail to achi...
This paper analyzes the topic identification stage of single-document automatic text summarization across four different domains, consisting of newswire, literary, scientific and ...
Hakan Ceylan, Rada Mihalcea, Umut O'zertem, Elena ...