Risk-controlling model selection via guided bayesian optimization

B Laufer-Goldshtein, A Fisch, R Barzilay… - arxiv preprint arxiv …, 2023‏ - arxiv.org
Adjustable hyperparameters of machine learning models typically impact various key trade-
offs such as accuracy, fairness, robustness, or inference cost. Our goal in this paper is to find …

Semi-Supervised Risk Control via Prediction-Powered Inference

BS Einbinder, L Ringel, Y Romano - arxiv preprint arxiv:2412.11174, 2024‏ - arxiv.org
The risk-controlling prediction sets (RCPS) framework is a general tool for transforming the
output of any machine learning model to design a predictive rule with rigorous error rate …

Fast yet Safe: Early-Exiting with Risk Control

M Jazbec, A Timans, TH Veljković, K Sakmann… - arxiv preprint arxiv …, 2024‏ - arxiv.org
Scaling machine learning models significantly improves their performance. However, such
gains come at the cost of inference being slow and resource-intensive. Early-exit neural …

Streamlining Conformal Information Retrieval via Score Refinement

Y Intrator, O Kelner, R Cohen, R Goldenberg… - arxiv preprint arxiv …, 2024‏ - arxiv.org
Information retrieval (IR) methods, like retrieval augmented generation, are fundamental to
modern applications but often lack statistical guarantees. Conformal prediction addresses …

Confident magnitude-based neural network pruning

J Alvarez - arxiv preprint arxiv:2408.04759, 2024‏ - arxiv.org
Pruning neural networks has proven to be a successful approach to increase the efficiency
and reduce the memory storage of deep learning models without compromising …