Search for: All records

Continuous-time dynamics models, e.g., neural ordinary differential equations, enable accurate modeling of underlying dynamics in time-series data. However, employing neural networks for parameterizing dynamics makes it challenging for humans to identify dependence structures, especially in the presence of delayed effects. In consequence, these models are not an attractive option when capturing dependence carries more importance than accurate modeling, e.g., in tsunami forecasting. In this paper, we present a novel method for identifying dependence structures in continuous-time dynamics models. We take a two-step approach: (1) During training, we promote weight sparsity in the model’s first layer during training. (2) We prune the sparse weights after training to identify dependence structures. In evaluation, we test our method in scenarios where the exact dependence structures of time-series are known. Compared to baselines, our method is more effective in uncovering dependence structures in data even when there are delayed effects. Moreover, we evaluate our method to a real-world tsunami forecasting, where the exact dependence structures are unknown beforehand. Even in this challenging scenario, our method still effective learns physically-consistent dependence structures and achieves high accuracy in forecasting. 

As event-based social networks (EBSNs) such as Meetup.com and Facebook Events gain popularity in managing local events (e.g., farmers’ markets and social gatherings), two-sided cultural niches are created as event organizers and participants benefit from the platform while affecting each other. Among various factors, niche overlap, an ecological feature, has been studied as a key factor that shapes the success of online communities. While such ecological factors may also shape EBSN-based local groups’ success, the context of EBSNs raises unique challenges in understanding the roles of cultural niches due to the informal nature of the local groups and their geographical embeddedness. In this paper, we examine the effects of Meetup groups’ topic overlap and geospatial correlation on the activity levels of both organizers and participants, using one-year Meetup data for 500 cities in the United States. We find that (1) a group’s topic overlap with other groups on EBSN is associated with its activity levels, and (2) local groups’ geospatial correlation may moderate the effects of topic overlap for EBSN users, but inconsistently. The results provide a baseline understanding of EBSN-based groups from an ecological perspective. 

Graph synthesis is a long-standing research problem. Many deep neural networks that learn about latent characteristics of graphs and generate fake graphs have been proposed. However, in many cases their scalability is too high to be used to synthesize large graphs. Recently, one work proposed an interesting scalable idea to learn and generate random walks that can be merged into a graph. Due to its difficulty, however, the random walk-based graph synthesis failed to show state-of-the-art performance in many cases. We present an improved random walk-based method by using negative random walks. In our experiments with 6 datasets and 8 baseline methods, our method shows the best performance in almost all cases. We achieve both high scalability and generation quality.