Optimised probabilistic active learning (OPAL)

by Georg Krempl, Daniel Kottke, Vincent Lemaire. In contrast to ever increasing volumes of automatically generated data, human annotation capacities remain limited. Thus, fast active learning approaches that allow the efficient allocation of annotation efforts gain in importance. Furthermore, cost-sensitive applications such as fraud detection pose the additional challenge of differing misclassification costs between classes. […]

Read More

Data-Driven Spine Detection for Multi-Sequence MRI

by Daniel Kottke, Gino Gulamhussene, Klaus Tönnies. Epidemiology studies on vertebra’s shape and appearance require big databases of medical images and image processing methods, that are robust against deformation and noise. This work presents a solution of the first step: the vertebrae detection. We propose a method that automatically detects the central spinal curve with […]

Read More

Probabilistic Active Learning: Towards Combining Versatility, Optimality and Efficiency

by Georg Krempl, Daniel Kottke Mining data with minimal annotation costs requires efficient active approaches, that ideally select the optimal candidate for labelling under a user-specified classification performance measure. Common generic approaches, that are usable with any classifier and any performance measure, are either slow like error reduction, or heuristics like uncertainty sampling. In contrast, […]

Read More

Probabilistic Active Learning: A Short Proposition

by Georg Krempl, Daniel Kottke, Myra Spiliopoulou. Active Mining of Big Data requires fast approaches that ideally select for a user-specified performance measure and arbitrary classifier the optimal instance for improving the classification performance. Existing generic approaches are either slow, like error reduction, or heuristics, like uncertainty sampling. We propose a novel, fast yet versatile […]

Read More