Our paper on accurately forecasting anomalous crowd dynamics has been published in International Journal “IEEE Pervasive Computing.”
When a large-scale event is held in a city, the technique of “early crowd dynamics forecasting,” which predicts the crowd density at the venue at an early stage, is very important as a means of crowd control or COVID-19 countermeasures. Conventional methods for this topic have been proposed that take into account external factors such as holidays and weather, as well as users’ future activity schedules [Konishi+ UbiComp “16; Anno+ SigSpatial’21].
However, events that induce congestion are rare and the small number of crowded patterns leads to an increase in model variance, which causes instability in prediction. To address this issue, we proposed Importance-based Synthetic Oversampling by extending the importance weighting proposed in [Anno+ SigSpatial’21] with Synthetic Minority Oversampling [Chawla+, Journal Of Artificial Intelligence Research 2002]. This method aims to reduce model variance by synthetically sampling more crowded patterns and increasing the number of learning patterns within a framework that theoretically guarantees the suppression of model bias.
We conducted the performance evaluation experiments similar to those in [Anno+ SigSpatial’21] and confirmed the predictive performance improvement of the proposed method.