Search

For a given set of spatial locations (e.g., criminal activities, disease outbreak, etc.), circular hotspot detection (CHD) identifies circular zones (or hotspots) of significantly high concentration of activity points in the given space. Furthermore, the prevalence of activities inside these zones is considerably higher than outside them. CHD is essential for numerous societal applications, including criminology, epidemiology, etc. The existing methods for CHD consider one of the given activity points as the center of the circular zone (or hotspot). Hence, these methods are incapable of ignoring small gaps (like mountains, rivers, etc.) in the spatial contiguity of hotspots and are not geographically robust. Geographically robust CHD (GR-CHD) requires enumeration of those candidate circles where the hotspot center is not necessarily an activity point. Unfortunately, this enumeration unreasonably increases the enumerated circles and makes GR-CHD a computationally expensive problem. Therefore, in this paper, GR-CHD is modeled as a single-objective optimization problem. Further, three swarm intelligence (SI) algorithms, namely particle swarm optimization, grey wolf optimizer, and salp swarm algorithm, are applied to the proposed model to detect geographically robust hotspots. Finally, the performances of the presented SI-based schemes and state-of-the-art cubic grid circle (CGC) algorithm have been compared. They are evaluated over the controlled synthetic datasets and the crime dataset of Chicago city for the year 2019. The obtained results indicate that the computational times required by the SI-based schemes are significantly less than the CGC algorithm. Also, the quality of hotspots detected by SI-based algorithms is either improved or at par with the CGC algorithm. [ABSTRACT FROM AUTHOR]

Accurate demand prediction of bike-sharing is an important prerequisite to reducing the cost of scheduling and improving the user satisfaction. However, it is a challenging issue due to stochasticity and non-linearity in bike-sharing systems. In this paper, a model called pseudo-double hidden layer feedforward neural networks is proposed to approximately predict actual demands of bike-sharing. Specifically, to overcome limitations in traditional back-propagation learning process, an algorithm, an extreme learning machine with improved particle swarm optimization, is designed to construct learning rules in neural networks. The performance is verified by comparing with other learning algorithms on the dataset of Streeter Dr bike-sharing station in Chicago. [ABSTRACT FROM AUTHOR]