Detecting and Mitigating Test-time Failure Risks via Model-agnostic Uncertainty Learning

@article{Lahoti2021DetectingAM,
  title={Detecting and Mitigating Test-time Failure Risks via Model-agnostic Uncertainty Learning},
  author={Preethi Lahoti and Krishna P. Gummadi and Gerhard Weikum},
  journal={2021 IEEE International Conference on Data Mining (ICDM)},
  year={2021},
  pages={1174-1179}
}
Reliably predicting potential failure risks of machine learning (ML) systems when deployed with production data is a crucial aspect of trustworthy AI. This paper introduces Risk Advisor, a novel post-hoc meta-learner for estimating failure risks and predictive uncertainties of any already-trained black-box classification model. In addition to providing a risk score, the Risk Advisor decomposes the uncertainty estimates into aleatoric and epistemic uncertainty components, thus giving informative… 
1 Citations

Figures and Tables from this paper

Responsible model deployment via model-agnostic uncertainty learning

The Risk Advisor is introduced, a novel post-hoc meta-learner for estimating failure risks and predictive uncertainties of any already-trained black-box classification model and decomposes the uncertainty estimates into aleatoric and epistemic uncertainty components, thus giving informative insights into the sources of uncertainty inducing the failures.

References

SHOWING 1-10 OF 38 REFERENCES

Learning to Validate the Predictions of Black Box Classifiers on Unseen Data

A simple approach to automate the validation of deployed ML models by estimating the model's predictive performance on unseen, unlabeled serving data and finds that it reliably predicts the performance of black box models in the majority of cases, and outperforms several baselines even in the presence of unspecified data errors.

Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift

A large-scale benchmark of existing state-of-the-art methods on classification problems and the effect of dataset shift on accuracy and calibration is presented, finding that traditional post-hoc calibration does indeed fall short, as do several other previous methods.

Can You Trust This Prediction? Auditing Pointwise Reliability After Learning

Resampling uncertainty estimation (RUE) is described, an algorithm to audit the pointwise reliability of predictions and can create predictive distributions that are competitive with state-of-the-art methods like Monte Carlo dropout, probabilistic backpropagation, and deep ensembles, but does not depend on specific algorithms at train-time like these methods do.

Decomposition of Uncertainty in Bayesian Deep Learning for Efficient and Risk-sensitive Learning

Using the decomposition of uncertainty in aleatoric and epistemic components for decision making purposes, a novel risk-sensitive criterion for reinforcement learning is defined to identify policies that balance expected cost, model-bias and noise aversion.

Reliable classification: Learning classifiers that distinguish aleatoric and epistemic uncertainty

Aleatoric and Epistemic Uncertainty with Random Forests

It is shown how two general approaches for measuring the learner's aleatoric and epistemic uncertainty in a prediction can be instantiated with decision trees and random forests as learning algorithms in a classification setting.

Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles

This work proposes an alternative to Bayesian NNs that is simple to implement, readily parallelizable, requires very little hyperparameter tuning, and yields high quality predictive uncertainty estimates.

Uncertainty in Gradient Boosting via Ensembles

Experiments on a range of regression and classification datasets show that ensembles of gradient boosting models yield improved predictive performance, and measures of uncertainty successfully enable detection of out-of-domain inputs.

Aleatoric and epistemic uncertainty in machine learning: an introduction to concepts and methods

This paper provides an introduction to the topic of uncertainty in machine learning as well as an overview of attempts so far at handling uncertainty in general and formalizing this distinction in particular.

Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning

A new theoretical framework is developed casting dropout training in deep neural networks (NNs) as approximate Bayesian inference in deep Gaussian processes, which mitigates the problem of representing uncertainty in deep learning without sacrificing either computational complexity or test accuracy.