Disproving the Feasibility of Learned Confidence Calibration Under Binary Supervision: An Information-Theoretic Impossibility

• URL: http://arxiv.org/abs/2509.14386v1
• Date: Wed, 17 Sep 2025 19:43:45 GMT
• Title: Disproving the Feasibility of Learned Confidence Calibration Under Binary Supervision: An Information-Theoretic Impossibility
• Abstract summary: neural networks cannot simultaneously learn well-calibrated confidence estimates with meaningful diversity when trained using correct/incorrect supervision.<n>We show this is an information-theoretic constraint, not a methodological failure.<n>We propose novel supervision paradigms using ensemble disagreement and adaptive multi-agent learning.
• Score: 2.1485350418225244
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We prove a fundamental impossibility theorem: neural networks cannot simultaneously learn well-calibrated confidence estimates with meaningful diversity when trained using binary correct/incorrect supervision. Through rigorous mathematical analysis and comprehensive empirical evaluation spanning negative reward training, symmetric loss functions, and post-hoc calibration methods, we demonstrate this is an information-theoretic constraint, not a methodological failure. Our experiments reveal universal failure patterns: negative rewards produce extreme underconfidence (ECE greater than 0.8) while destroying confidence diversity (std less than 0.05), symmetric losses fail to escape binary signal averaging, and post-hoc methods achieve calibration (ECE less than 0.02) only by compressing the confidence distribution. We formalize this as an underspecified mapping problem where binary signals cannot distinguish between different confidence levels for correct predictions: a 60 percent confident correct answer receives identical supervision to a 90 percent confident one. Crucially, our real-world validation shows 100 percent failure rate for all training methods across MNIST, Fashion-MNIST, and CIFAR-10, while post-hoc calibration's 33 percent success rate paradoxically confirms our theorem by achieving calibration through transformation rather than learning. This impossibility directly explains neural network hallucinations and establishes why post-hoc calibration is mathematically necessary, not merely convenient. We propose novel supervision paradigms using ensemble disagreement and adaptive multi-agent learning that could overcome these fundamental limitations without requiring human confidence annotations.

Related papers

• DirEAG: Dirichlet Evidence Aggregation for Calibrating Verbalized Confidence in Mathematical Reasoning [2.867517731896504]
  Black-box verbalized confidence is difficult to calibrate in large language models.<n>We propose DirEAG, a Dirichlet Evidence Aggregation method that converts each elicited answer-confidence observation into calibrated soft evidence.
  arXiv  Detail & Related papers  (2026-08-21T03:48:25Z)
• From token probabilities to calibrated confidence: An empirical study of mathematical question answering [1.0032961794537367]
  Prior work has shown that token probabilities are often overconfident.<n>We compare single-pass estimators, which reuse token probabilities from the original generation, with multi-pass estimators.<n>We find that aggregating token probabilities over the full sequence captures small but consistent differences between correct and incorrect generations.
  arXiv  Detail & Related papers  (2026-08-08T00:00:19Z)
• Process Supervision of Confidence Margin for Calibrated LLM Reasoning [52.373121066425455]
  Scaling test-time computation with reinforcement learning (RL) has emerged as a reliable path to improve large language models (LLM) reasoning ability.<n>Yet, outcome-based reward often incentivizes models to be overconfident, leading to hallucinations, unreliable confidence-based control, and unnecessary compute allocation.<n>We introduce Reinforcement Learning with Confidence Margin (textbfRLCM), a calibration-aware RL framework that jointly optimize correctness and confidence reliability.
  arXiv  Detail & Related papers  (2026-04-25T14:40:13Z)
• Socrates Loss: Unifying Confidence Calibration and Classification by Leveraging the Unknown [4.503897051444244]
  Deep neural networks, despite their high accuracy, often exhibit poor confidence calibration, limiting their reliability in high-stakes applications.<n>Current ad-hoc confidence calibration methods attempt to fix this during training but face a fundamental trade-off.<n>This paper addresses and mitigates this stability-performance trade-off.<n>We propose Socrates Loss, a novel, unified loss function that explicitly leverages uncertainty by incorporating an auxiliary unknown class.
  arXiv  Detail & Related papers  (2026-04-14T03:43:15Z)
• Know When You're Wrong: Aligning Confidence with Correctness for LLM Error Detection [0.0]
  Large language models (LLMs) are increasingly deployed in critical decision-making systems.<n>We introduce a normalized confidence score based on output anchor token probabilities.<n>This enables direct detection of errors and hallucinations with minimal overhead.
  arXiv  Detail & Related papers  (2026-02-18T07:05:12Z)
• Geometric Calibration and Neutral Zones for Uncertainty-Aware Multi-Class Classification [0.0]
  This work bridges information geometry and statistical learning, offering formal guarantees for uncertainty-aware classification in applications requiring rigorous validation.<n> Empirical validation on Adeno-Associated Virus classification demonstrates that the two-stage framework captures 72.5% of errors while deferring 34.5% of samples, reducing automated decision error rates from 16.8% to 6.9%.
  arXiv  Detail & Related papers  (2025-11-26T01:29:49Z)
• Annotation-Efficient Universal Honesty Alignment [70.05453324928955]
  Existing methods either rely on training-free confidence estimation or training-based calibration with correctness annotations.<n>Elicitation-Then-Calibration (EliCal) is a two-stage framework that first elicits internal confidence using inexpensive self-consistency supervision, then calibrates this confidence with a small set of correctness annotations.<n>EliCal achieves near-optimal alignment with only 1k correctness annotations (0.18% of full supervision) and better alignment performance on unseen MMLU tasks than the calibration-only baseline.
  arXiv  Detail & Related papers  (2025-10-20T13:05:22Z)
• Uncertainty-Aware Post-Hoc Calibration: Mitigating Confidently Incorrect Predictions Beyond Calibration Metrics [6.9681910774977815]
  This paper presents a post-hoc calibration framework to enhance calibration quality and uncertainty-aware decision-making.<n>A comprehensive evaluation is conducted using calibration metrics, uncertainty-aware performance measures, and empirical conformal coverage.<n> Experiments show that the proposed method achieves lower confidently incorrect predictions, and competitive Expected Error compared with isotonic and focal-loss baselines.
  arXiv  Detail & Related papers  (2025-10-19T23:55:36Z)
• ConfTuner: Training Large Language Models to Express Their Confidence Verbally [58.63318088243125]
  Large Language Models (LLMs) are increasingly deployed in high-stakes domains such as science, law, and healthcare.<n>LLMs are often observed to generate incorrect answers with high confidence, a phenomenon known as "overconfidence"
  arXiv  Detail & Related papers  (2025-08-26T09:25:32Z)
• CLUE: Neural Networks Calibration via Learning Uncertainty-Error alignment [7.702016079410588]
  We introduce CLUE (Calibration via Learning Uncertainty-Error Alignment), a novel approach that aligns predicted uncertainty with observed error during training.<n>We show that CLUE achieves superior calibration quality and competitive predictive performance with respect to state-of-the-art approaches.
  arXiv  Detail & Related papers  (2025-05-28T19:23:47Z)
• Enhance GNNs with Reliable Confidence Estimation via Adversarial Calibration Learning [30.450482094196243]
  Despite their impressive predictive performance, GNNs often exhibit poor confidence calibration.<n>This issue raises concerns about their reliability in high-stakes domains such as fraud detection, and risk assessment.<n>We propose a novel AdvCali framework that adaptively enhances calibration across different node groups.
  arXiv  Detail & Related papers  (2025-03-23T23:04:41Z)
• Pretraining with random noise for uncertainty calibration [0.0]
  Many models exhibit hallucinations - confident yet inaccurate responses - due to miscalibrated confidence.<n>Our method, inspired by developmental neuroscience, addresses this issue by simply pretraining networks with random noise and labels.
  arXiv  Detail & Related papers  (2024-12-23T09:22:00Z)
• Binary Classification with Confidence Difference [100.08818204756093]
  This paper delves into a novel weakly supervised binary classification problem called confidence-difference (ConfDiff) classification. We propose a risk-consistent approach to tackle this problem and show that the estimation error bound the optimal convergence rate. We also introduce a risk correction approach to mitigate overfitting problems, whose consistency and convergence rate are also proven.
  arXiv  Detail & Related papers  (2023-10-09T11:44:50Z)
• Two Sides of Miscalibration: Identifying Over and Under-Confidence Prediction for Network Calibration [1.192436948211501]
  Proper confidence calibration of deep neural networks is essential for reliable predictions in safety-critical tasks. Miscalibration can lead to model over-confidence and/or under-confidence. We introduce a novel metric, a miscalibration score, to identify the overall and class-wise calibration status. We use the class-wise miscalibration score as a proxy to design a calibration technique that can tackle both over and under-confidence.
  arXiv  Detail & Related papers  (2023-08-06T17:59:14Z)
• Beyond calibration: estimating the grouping loss of modern neural networks [68.8204255655161]
  Proper scoring rule theory shows that given the calibration loss, the missing piece to characterize individual errors is the grouping loss. We show that modern neural network architectures in vision and NLP exhibit grouping loss, notably in distribution shifts settings.
  arXiv  Detail & Related papers  (2022-10-28T07:04:20Z)
• Adversarial Training with Rectified Rejection [114.83821848791206]
  We propose to use true confidence (T-Con) as a certainty oracle, and learn to predict T-Con by rectifying confidence. We prove that under mild conditions, a rectified confidence (R-Con) rejector and a confidence rejector can be coupled to distinguish any wrongly classified input from correctly classified ones.
  arXiv  Detail & Related papers  (2021-05-31T08:24:53Z)
• Don't Just Blame Over-parametrization for Over-confidence: Theoretical Analysis of Calibration in Binary Classification [58.03725169462616]
  We show theoretically that over-parametrization is not the only reason for over-confidence. We prove that logistic regression is inherently over-confident, in the realizable, under-parametrized setting. Perhaps surprisingly, we also show that over-confidence is not always the case.
  arXiv  Detail & Related papers  (2021-02-15T21:38:09Z)
• Binary Classification from Positive Data with Skewed Confidence [85.18941440826309]
  Positive-confidence (Pconf) classification is a promising weakly-supervised learning method. In practice, the confidence may be skewed by bias arising in an annotation process. We introduce the parameterized model of the skewed confidence, and propose the method for selecting the hyper parameter.
  arXiv  Detail & Related papers  (2020-01-29T00:04:36Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.