Learning Optimal Conformal Classifiers

2021/10/18 by David Stutz, Stutz, David, Krishnamurthy +5 · 26 citations
Computer Science · Mathematics · #Adversarial Robustness in Machine Learning #Computer Vision and Pattern Recognition (cs.CV) #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Machine Learning and Data Classification #Methodology (stat.ME) #cs.CV #cs.LG #stat.ME #stat.ML

paper · pdf · doi:10.48550/arxiv.2110.09192

ICLR 2022

openalex publication_date 2021/10/18 · arxiv created 2022/05/06 · arxiv updated 2022/05/09 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Modern deep learning based classifiers show very high accuracy on test data but this does not provide sufficient guarantees for safe deployment, especially in high-stake AI applications such as medical diagnosis. Usually, predictions are obtained without a reliable uncertainty estimate or a formal guarantee. Conformal prediction (CP) addresses these issues by using the classifier's predictions, e.g., its probability estimates, to predict confidence sets containing the true class with a user-specified probability. However, using CP as a separate processing step after training prevents the underlying model from adapting to the prediction of confidence sets. Thus, this paper explores strategies to differentiate through CP during training with the goal of training model with the conformal wrapper end-to-end. In our approach, conformal training (ConfTr), we specifically "simulate" conformalization on mini-batches during training. Compared to standard training, ConfTr reduces the average confidence set size (inefficiency) of state-of-the-art CP methods applied after training. Moreover, it allows to "shape" the confidence sets predicted at test time, which is difficult for standard CP. On experiments with several datasets, we show ConfTr can influence how inefficiency is distributed across classes, or guide the composition of confidence sets in terms of the included classes, while retaining the guarantees offered by CP.

Citations

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Gradient-based learning applied to document recognition
• Least Ambiguous Set-Valued Classifiers with Bounded Error Levels
• EMNIST: an extension of MNIST to handwritten letters
• Improved Regularization of Convolutional Neural Networks with Cutout
• Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
• AutoAugment: Learning Augmentation Policies from Data
• International evaluation of an AI system for breast cancer screening
• Learning Multiple Layers of Features from Tiny Images
• Distribution-Free, Risk-Controlling Prediction Sets

Cited by

• Conformal Correction for Efficiency May be at Odds with Entropy
• CLAPS: Aleatoric-Epistemic Scaling via Last-Layer Laplace for Conformal Regression
• Cost-Sensitive Conformal Training with Provably Controllable Learning Bounds
• Elements of Conformal Prediction for Statisticians
• CoFineLLM: Conformal Finetuning of LLMs for Language-Instructed Robot Planning
• Conformal Changepoint Localization and Root Cause Analysis with Corrupted Observations
• Conformal Risk Training: End-to-End Optimization of Conformal Risk Control
• Annotation-Efficient Active Test-Time Adaptation with Conformal Prediction
• Conformal Prediction for Signal Temporal Logic Inference
• Singleton-Optimized Conformal Prediction
• Smoothing-Based Conformal Prediction for Balancing Efficiency and Interpretability
• Learning Conformal Explainers for Image Classifiers
• Deceptive Risk Minimization: Out-of-Distribution Generalization by Deceiving Distribution Shift Detectors
• Optimizing In-Context Learning for Efficient Full Conformal Prediction
• Conformal Prediction and Trustworthy AI
• Non-exchangeable Conformal Prediction with Optimal Transport: Tackling Distribution Shifts with Unlabeled Data
• Enhancing Adversarial Robustness with Conformal Prediction: A Framework for Guaranteed Model Reliability
• Direct Prediction Set Minimization via Bilevel Conformal Classifier Training
• Full Conformal Adaptation of Medical Vision-Language Models
• Conformal Prediction for Zero-Shot Models
• Dynamic Estimation Loss Control in Variational Quantum Sensing via Online Conformal Inference
• Individualised Counterfactual Examples Using Conformal Prediction Intervals
• Risk-Sensitive Conformal Prediction for Catheter Placement Detection in Chest X-rays
• ASAT: Adaptive Scoring and Thresholding with Human Feedback for Robust Out-of-Distribution Detection
• Backward Conformal Prediction
• Conformal Policy Control

Related

• C-Adapter: Adapting Deep Classifiers for Efficient Conformal Prediction Sets
• Direct Prediction Set Minimization via Bilevel Conformal Classifier Training
• Test-time augmentation improves efficiency in conformal prediction
• Provably Robust Conformal Prediction with Improved Efficiency
• Cost-Sensitive Conformal Training with Provably Controllable Learning Bounds