This study examines the impact of uncertainty in the land use component of a partially integrated land use-transportation modeling system called UrbanSim. Outputs from the land use model (LUM) act as inputs for a traditional 4-step travel demand model (TDM), and travel times from the traffic-assignment stage of the TDM are fed forward into the subsequent years LUM. This work examines the propagation of uncertainty across model stages as well as at each model stage over time. A factorized design approach is used to model uncertainty in demographic inputs (which include aggregate growth rates and mobility rates) to the land use model, as well as uncertainty in various model parameters. The results suggest that while several model inputs may affect model outputs in the short run, only those inputs that have a cumulative effect are likely to have a significant impact on outputs in the long run. The results also suggest that uncertainty in model outputs may increase for the first few years for which the model is run, as modified inputs send shocks through the urban system. However, the level of uncertainty appears to come down in later years, as households, jobs, and developers respond to changed input conditions KeywordsError propagation, uncertainty representation, integrated models, land use, travel demand model 2

Highlights A U-Net incorporating spatial prior information can successfully segment 6 brain tissue types. The U-Net was able to segment gray and white matter in the presence of lesions. The U-Net surpassed the performance of its source algorithm in an external dataset. Segmentations were produced in a hundredth of the time of its predecessor algorithm.

No abstract

Document-level information extraction is a flexible framework compatible with applications where information is not necessarily localized in a single sentence. For example, key features of a diagnosis in radiology a report may not be explicitly stated, but nevertheless can be inferred from the report's text. However, document-level neural models can easily learn spurious correlations from irrelevant information. This work studies how to ensure that these models make correct inferences from complex text and make those inferences in an auditable way: beyond just being right, are these models "right for the right reasons?" We experiment with post-hoc evidence extraction in a predict-select-verify framework using feature attribution techniques. While this basic approach can extract reasonable evidence, it can be regularized with small amounts of evidence supervision during training, which substantially improves the quality of extracted evidence. We evaluate on two domains: a small-scale labeled dataset of brain MRI reports and a large-scale modified version of Do-cRED (Yao et al., 2019) and show that models' plausibility can be improved with no loss in accuracy.