Two out of Three (ToT): using self-consistency to make robust predictions

2025-05-19

AI Summary

Deep learning models face deployment risks in high-stakes applications due to opaque decision-making and poorly quantified uncertainty. To address this, we propose a training-free, unsupervised, model-agnostic abstention mechanism that integrates uncertainty awareness via prompt engineering and multi-path sampling. It generates the original prediction alongside two alternative predictions, forming a three-way self-consistency criterion. Crucially, we formalize human sensitivity to conflicting evidence as a joint gating mechanism combining majority voting and confidence divergence—dynamically triggering abstention when consensus is weak. Evaluated on ImageNet, out-of-distribution detection, and medical diagnosis benchmarks, our method reduces average error rate by 37%, achieves task coverage exceeding 92%, and attains 94.6% abstention accuracy. It significantly enhances robustness and safety in safety-critical domains without requiring model retraining or access to ground-truth uncertainty labels.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations
• Natural Language Processing: Safety and Robustness

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Deep learning (DL) can automatically construct intelligent agents, deep neural networks (alternatively, DL models), that can outperform humans in certain tasks. However, the operating principles of DL remain poorly understood, making its decisions incomprehensible. As a result, it poses a great risk to deploy DL in high-stakes domains in which mistakes or errors may lead to critical consequences. Here, we aim to develop an algorithm that can help DL models make more robust decisions by allowing them to abstain from answering when they are uncertain. Our algorithm, named `Two out of Three (ToT)', is inspired by the sensitivity of the human brain to conflicting information. ToT creates two alternative predictions in addition to the original model prediction and uses the alternative predictions to decide whether it should provide an answer or not.

Problem

• Enhancing DL model decision robustness via uncertainty abstention
• Addressing incomprehensible DL decisions in high-stakes domains
• Proposing ToT algorithm for self-consistent conflict-sensitive predictions

Innovation

• Uses self-consistency for robust predictions
• Creates two alternative predictions for validation
• Allows abstention when predictions conflict