# Human uncertainty makes classification more robust

arXiv: https://arxiv.org/abs/1908.07086 · HTML: https://mlsanity.com/paper/1908.07086/

## Key facts

- Cited by 231 papers in the map (data through 7 Oct 2026).
- First posted on arXiv 19 Aug 2019.

## Built on by (most cited first, 30 of 231)

- [Learning with Noisy Labels Revisited: A Study Using Real-World Human Annotations](https://mlsanity.com/paper/2110.12088/) (2021)
- [Are We Done with MMLU?](https://mlsanity.com/paper/2406.04127/) (2024)
- [Getting aligned on representational alignment](https://mlsanity.com/paper/2310.13018/) (2023)
- [The 'Problem' of Human Label Variation: On Ground Truth in Data, Modeling and Evaluation](https://mlsanity.com/paper/2211.02570/) (2022)
- [Toward a Perspectivist Turn in Ground Truthing for Predictive Computing](https://mlsanity.com/paper/2109.04270/) (2021)
- [Jury Learning: Integrating Dissenting Voices into Machine Learning Models](https://mlsanity.com/paper/2202.02950/) (2022)
- [Plex: Towards Reliability using Pretrained Large Model Extensions](https://mlsanity.com/paper/2207.07411/) (2022)
- [DataPerf: Benchmarks for Data-Centric AI Development](https://mlsanity.com/paper/2207.10062/) (2022)
- [Benchmarking Uncertainty Disentanglement: Specialized Uncertainties for Specialized Tasks](https://mlsanity.com/paper/2402.19460/) (2024)
- [Agreeing to Disagree: Annotating Offensive Language Datasets with Annotators' Disagreement](https://mlsanity.com/paper/2109.13563/) (2021)
- [Improving Expert Predictions with Conformal Prediction](https://mlsanity.com/paper/2201.12006/) (2022)
- [SemEval-2023 Task 11: Learning With Disagreements (LeWiDi)](https://mlsanity.com/paper/2304.14803/) (2023)
- [Active label cleaning for improved dataset quality under resource constraints](https://mlsanity.com/paper/2109.00574/) (2021)
- [Eliciting and Learning with Soft Labels from Every Annotator](https://mlsanity.com/paper/2207.00810/) (2022)
- [Is one annotation enough? A data-centric image classification benchmark for noisy and ambiguous label estimation](https://mlsanity.com/paper/2207.06214/) (2022)
- [VariErr NLI: Separating Annotation Error from Human Label Variation](https://mlsanity.com/paper/2403.01931/) (2024)
- [Combining Human Predictions with Model Probabilities via Confusion Matrices and Calibration](https://mlsanity.com/paper/2109.14591/) (2021)
- [Forming Effective Human-AI Teams: Building Machine Learning Models that Complement the Capabilities of Multiple Experts](https://mlsanity.com/paper/2206.07948/) (2022)
- [CLIP in Medical Imaging: A Survey](https://mlsanity.com/paper/2312.07353/) (2023)
- [Human Uncertainty in Concept-Based AI Systems](https://mlsanity.com/paper/2303.12872/) (2023)
- [AmbiFC: Fact-Checking Ambiguous Claims with Evidence](https://mlsanity.com/paper/2104.00640/) (2021)
- [Improving neural network representations using human similarity judgments](https://mlsanity.com/paper/2306.04507/) (2023)
- [Data augmentation in Bayesian neural networks and the cold posterior effect](https://mlsanity.com/paper/2106.05586/) (2021)
- [Trivial or impossible -- dichotomous data difficulty masks model differences (on ImageNet and beyond)](https://mlsanity.com/paper/2110.05922/) (2021)
- [On the Utility of Prediction Sets in Human-AI Teams](https://mlsanity.com/paper/2205.01411/) (2022)
- [Distributed NLI: Learning to Predict Human Opinion Distributions for Language Reasoning](https://mlsanity.com/paper/2104.08676/) (2021)
- [Conformal prediction under ambiguous ground truth](https://mlsanity.com/paper/2307.09302/) (2023)
- [To Aggregate or Not? Learning with Separate Noisy Labels](https://mlsanity.com/paper/2206.07181/) (2022)
- [Probabilistic Contrastive Learning Recovers the Correct Aleatoric Uncertainty of Ambiguous Inputs](https://mlsanity.com/paper/2302.02865/) (2023)
- [Human-Aided Saliency Maps Improve Generalization of Deep Learning](https://mlsanity.com/paper/2105.03492/) (2021)

## Abstract (arXiv)

The classification performance of deep neural networks has begun to asymptote at near-perfect levels. However, their ability to generalize outside the training set and their robustness to adversarial attacks have not. In this paper, we make progress on this problem by training with full label distributions that reflect human perceptual uncertainty. We first present a new benchmark dataset which we call CIFAR10H, containing a full distribution of human labels for each image of the CIFAR10 test set. We then show that, while contemporary classifiers fail to exhibit human-like uncertainty on their own, explicit training on our dataset closes this gap, supports improved generalization to increasingly out-of-training-distribution test datasets, and confers robustness to adversarial attacks.