Permafrost thaw is an important aspect of Earth’s carbon cycles. Initial estimates suggest that permafrost thaw could contribute anywhere between 20 to 500 Gt of CO2-eq by 2100. These estimates are all fundamentally centered around one number: active layer thickness (ALT). The deeper the ALT, the more emissions. Unfortunately, ALT is a highly spatially heterogeneous number and determined by numerous thermal, soil hydrology, and geomorphological effects. The remotely sensed active layer thickness (ReSALT) algorithm was introduced in 2010 to provide scientists with a way to model ALT heterogeneously at meter-scale resolution. However, upon inspection, this work shows that ReSALT’s modeling approach is self-inconsistent when using a variable soil porosity model. This work then introduces SCReSALT (Self-Consistent Re-SALT) to solve that problem. Experimental comparisons to a past study in Utqiagvik (formerly Barrow), Alaska show significant improvement and suggest that SCReSALT should become a standard tool in permafrost thaw modeling. Code to reproduce all results can be found at https://github.com/jmathur25/permafrost-prediction.

Metric elicitation is a recent framework for eliciting classification performance metrics that best reflect implicit user preferences based on the task and context. However, available elicitation strategies have been limited to linear (or quasi-linear) functions of predictive rates, which can be practically restrictive for many applications including fairness. This paper develops a strategy for eliciting more flexible multiclass metrics defined by quadratic functions of rates, designed to reflect human preferences better. We show its application in eliciting quadratic violation-based group-fair metrics. Our strategy requires only relative preference feedback, is robust to noise, and achieves near-optimal query complexity. We further extend this strategy to eliciting polynomial metrics – thus broadening the use cases for metric elicitation.

We consider learning to optimize a classification metric defined by a black-box function of the confusion matrix. Such black-box learning settings are ubiquitous, for example, when the learner only has query access to the metric of interest, or in noisy-label and domain adaptation applications where the learner must evaluate the metric via performance evaluation using a small validation sample. Our approach is to adaptively learn example weights on the training dataset such that the resulting weighted objective best approximates the metric on the validation sample. We show how to model and estimate the example weights and use them to iteratively post-shift a pre-trained class probability estimator to construct a classifier. We also analyze the resulting procedure’s statistical properties. Experiments on various label noise, domain shift, and fair classification setups confirm that our proposal compares favorably to the state-of-the-art baselines for each application.