We investigate a novel cluster-of-bandit algorithm CAB for collaborative recommendation tasks that implements the underlying feedback sharing mechanism by estimating user neighborhoods in a context-dependent manner. CAB makes sharp departures from the state of the art by incorporating collaborative effects into inference, as well as learning processes in a manner that seamlessly interleaves explore-exploit tradeoffs and collaborative steps. We prove regret bounds for CAB under various data-dependent assumptions which exhibit a crisp dependence on the expected number of clusters over the users, a natural measure of the statistical difficulty of the learning task. Experiments on production and real-world datasets show that CAB offers significantly increased prediction performance against a representative pool of state-of-the-art methods.

Among the most serious natural disasters, earthquakes cause severe damages to infrastructures and building, can kill or injure thousands of humans and animals and, in the luckiest circumstances, just make people homeless destroying communities, habitats, economies and mental equilibrium. In order to minimise the loss of lives, an effective evacuation plan to cope with worldwide disasters is required. In this paper we describe a novel approach to timely formulate an evacuation plan of an area struck by an earthquake. The proposed solution leverages on a two-steps modeling framework: i) a method that extracts from enriched GIS data a network description of the area to be evacuated; ii) a dynamic optimization model that calculates the safest paths citizens should follow to reach pre-identified safe areas. While the network is computed off-line at design time, the optimization model, or one of its reductions, can be embedded in a real-time system that, recomputing it several times, can guide citizen after a natural disaster even in case of high dynamic scenario. Our approach is demonstrated on a real study case: the medieval center of the Italian town of Sulmona, for which detailed GIS data with information on the urban structure and building vulnerability are available.

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis posuere tellus ac convallis placerat. Proin tincidunt magna sed ex sollicitudin condimentum.