Search

Social media messages posted by people during natural disasters often contain important location descriptions, such as the locations of victims. Recent research has shown that many of these location descriptions go beyond simple place names, such as city names and street names, and are difficult to extract using typical named entity recognition (NER) tools. While advanced machine learning models could be trained, they require large labeled training datasets that can be time-consuming and labor-intensive to create. In this work, we propose a method that fuses geo-knowledge of location descriptions and a Generative Pre-trained Transformer (GPT) model, such as ChatGPT and GPT-4. The result is a geo-knowledge-guided GPT model that can accurately extract location descriptions from disaster-related social media messages. Also, only 22 training examples encoding geo-knowledge are used in our method. We conduct experiments to compare this method with nine alternative approaches on a dataset of tweets from Hurricane Harvey. Our method demonstrates an over 40% improvement over typically used NER approaches. The experiment results also show that geo-knowledge is indispensable for guiding the behavior of GPT models. The extracted location descriptions can help disaster responders reach victims more quickly and may even save lives. [ABSTRACT FROM AUTHOR]

Predictive simulations of two-phase flows are highly sought after because of their widespread applications in propulsion, energy, agriculture, and medicine. One crucial goal for many of these simulations is the accurate and efficient prediction of the size distribution and number density of atomized drops. The multi-scale nature of these flows makes it practically impossible to capture all scales within a single simulation. In particular, the breakup processes producing the smallest drops through secondary breakup often necessitate resolutions far below the Kolmogorov scale. Consequently, models must be employed for secondary breakup. Existing physics-based and stochastic breakup models are not universal and fail to account for the local and instantaneous flow field and drop geometry. We present a physics-informed machine learning model for predicting the statistics of daughter drops generated during the breakup of under-resolved drops. By training on high-fidelity simulations, the model can predict breakup outcomes from severely under-resolved input fields. This is made possible by a careful choice of quantities of interest and by taking inspiration from the discrete nature of breakup events to encode the temporal evolution via a mixture of sigmoid functions. We showcase proof-of-concept results from the canonical settings of 3D Taylor–Green vortex flows and homogeneous isotropic turbulence. Compared to results generated by low-resolution simulations (i.e., without a model) and baseline state-of-the-art models, our approach achieves superior accuracy in predicting drop size distribution and critical quantities of interest, such as surface area. This work paves the path towards the development of embedded machine learning models for the prediction of primary and secondary breakup. [Display omitted] • Physics-informed ML model for statistics of daughter drops in under-resolved breakup • Found a suitable combination of models, quantities of interest, and loss functions • Encoded the time-dependence of the statistics using a mixture of sigmoids model • Enforced physical constraints such as mass conservation through the loss function • Demonstrated our ML model's accuracy in comparison to alternative baseline models [ABSTRACT FROM AUTHOR]