arXiv subjects

Kerri Lu

Publications and source records attributed to Kerri Lu.

3 recordsLinked to original sources

arXiv2026

Conformal Prediction Sets for Instance Segmentation

Current instance segmentation models achieve high performance on average predictions, but lack principled uncertainty quantification: their outputs are not calibrated, and there is no guarantee that a predicted mask is close to the ground truth. To address this limitation, we introduce a conformal prediction algorithm to generate adaptive confidence sets for instance segmentation. Given an image and a pixel coordinate query, our algorithm generates a confidence set of instance predictions for that pixel, with a provable guarantee for the probability that at least one of the predictions has high Intersection-Over-Union (IoU) with the true object instance mask. We apply our algorithm to instance segmentation examples in agricultural field delineation, cell segmentation, and vehicle detection. Empirically, we find that our prediction sets vary in size based on query difficulty and attain the target coverage, outperforming baselines (naive best parameter and morphological dilation-based methods). We provide versions of the algorithm with asymptotic and finite sample guarantees. Our work is the first to capture structural uncertainty in instance segmentation by constructing confidence sets of diverse segmentation predictions.

cs.CV↗

arXiv2025

Prediction-Powered Inference with Imputed Covariates and Nonuniform Sampling

Machine learning models are increasingly used to produce predictions that serve as input data in subsequent statistical analyses. For example, computer vision predictions of economic and environmental indicators based on satellite imagery are used in downstream regressions; similarly, language models are widely used to approximate human ratings and opinions in social science research. However, failure to properly account for errors in the machine learning predictions renders standard statistical procedures invalid. Prior work uses what we call the Predict-Then-Debias estimator to give valid confidence intervals when machine learning algorithms impute missing variables, assuming a small complete sample from the population of interest. We expand the scope by introducing bootstrap confidence intervals that apply when the complete data is a nonuniform (i.e., weighted, stratified, or clustered) sample and to settings where an arbitrary subset of features is imputed. Importantly, the method can be applied to many settings without requiring additional calculations. We prove that these confidence intervals are valid under no assumptions on the quality of the machine learning model and are no wider than the intervals obtained by methods that do not use machine learning predictions.

stat.ME↗

arXiv2025

Regression coefficient estimation from remote sensing maps

Regressions are commonly used in environmental science and economics to identify causal or associative relationships between variables. In these settings, remote sensing-derived map products increasingly serve as sources of variables, enabling estimation of effects such as the impact of conservation zones on deforestation. However, the quality of map products varies, and -- because maps are outputs of complex machine learning algorithms that take in a variety of remotely sensed variables as inputs -- errors are difficult to characterize. Thus, population-level estimators from such maps may be biased. In this paper, we apply prediction-powered inference (PPI) to estimate regression coefficients relating a response variable and covariates to each other. PPI is a method that estimates parameters of interest by using a small amount of randomly sampled ground truth data to correct for bias in large-scale remote sensing map products. Applying PPI across multiple remote sensing use cases in regression coefficient estimation, we find that it results in estimates that are (1) more reliable than using the map product as if it were 100% accurate and (2) have lower uncertainty than using only the ground truth sample data and ignoring the map product. Empirically, we observe effective sample size increases of up to 17-fold using PPI compared to only using ground truth data. This is the first work to estimate remote sensing regression coefficients without assumptions on the structure of map product errors. Data and code are available at https://github.com/Earth-Intelligence-Lab/uncertainty-quantification.

stat.AP↗