Uncertainty Quantification 360: A Holistic Toolkit for Quantifying and Communicating the Uncertainty of AI

In this paper, we describe an open source Python toolkit named Uncertainty Quantification 360 (UQ360) for the uncertainty quantification of AI models. The goal of this toolkit is twofold: first, to provide a broad range of capabilities to streamline as well as foster the common practices of quantifying, evaluating, improving, and communicating uncertainty in the AI application development lifecycle; second, to encourage further exploration of UQ’s connections to other pillars of trustworthy AI such as fairness and transparency through the dissemination of latest research and education materials. Beyond the Python package (https://github.com/IBM/UQ360), we have developed an interactive experience (http: //uq360.mybluemix.net) and guidance materials as educational tools to aid researchers and developers in producing and communicating high-quality uncertainties in an effective manner.

[1]  Yunfeng Zhang,et al.  Effect of confidence and explanation on accuracy and trust calibration in AI-assisted decision making , 2020, FAT*.

[2]  Milos Hauskrecht,et al.  Obtaining Well Calibrated Probabilities Using Bayesian Binning , 2015, AAAI.

[3]  Soumya Ghosh,et al.  Approximate Cross-Validation for Structured Models , 2020, NeurIPS.

[4]  Carl E. Rasmussen,et al.  Gaussian processes for machine learning , 2005, Adaptive computation and machine learning.

[5]  Soumya Ghosh,et al.  Uncertainty Characteristics Curves: A Systematic Assessment of Prediction Intervals , 2021, ArXiv.

[6]  Jirí Navrátil,et al.  Uncertainty Prediction for Deep Sequential Regression Using Meta Models , 2020, ArXiv.

[7]  Sean A. Munson,et al.  Uncertainty Displays Using Quantile Dotplots or CDFs Improve Transit Decision-Making , 2018, CHI.

[8]  Bianca Zadrozny,et al.  Obtaining calibrated probability estimates from decision trees and naive Bayesian classifiers , 2001, ICML.

[9]  Jon Wakefield,et al.  Bayesian and Frequentist Regression Methods , 2013 .

[10]  Matthew Arnold,et al.  Learning Prediction Intervals for Model Performance , 2020, AAAI.

[11]  G. Brier.  VERIFICATION OF FORECASTS EXPRESSED IN TERMS OF PROBABILITY , 1950 .

[12]  Soumya Ghosh,et al.  Model Selection in Bayesian Neural Networks via Horseshoe Priors , 2017, J. Mach. Learn. Res..

[13]  Durga L. Shrestha,et al.  Machine learning approaches for estimation of prediction interval for the model output , 2006, Neural Networks.

[14]  Chris Chatfield,et al.  Calculating Interval Forecasts , 1993 .

[15]  Stephen E. Fienberg,et al.  The Comparison and Evaluation of Forecasters. , 1983 .

[16]  Peer-Timo Bremer,et al.  Building Calibrated Deep Models via Uncertainty Matching with Auxiliary Interval Predictors , 2020, AAAI.

[17]  Karthikeyan Shanmugam,et al.  Confidence Scoring Using Whitebox Meta-models with Linear Classifier Probes , 2018, AISTATS.

[18]  Ran El-Yaniv,et al.  On the Foundations of Noise-free Selective Classification , 2010, J. Mach. Learn. Res..

[19]  Rumi Chunara,et al.  Uncertainty as a Form of Transparency: Measuring, Communicating, and Using Uncertainty , 2020, AIES.

[20]  Julien Cornebise,et al.  Weight Uncertainty in Neural Network , 2015, ICML.