ViCE: Virtual Conference on Epistemic Uncertainty in Engineering

This event encourages discussion and collaborations on modeling and decision making in the face of epistemic and aleatory uncertainty in practical engineering contexts. Building on the recent REC meeting in Liverpool and the RUC meetings in Cambridge and Amsterdam, we hope that this event will feature even more discussion and debate, serving as a workshop to address and perhaps come to concrete shared conclusions about handling uncertainty in engineering, and these questions:

• Decision making: Must decisions and designs distinguish kinds of uncertainties?

• Uncertainty arithmetic: Do epistemic and aleatory need different calculi?

• Uncertainty in engineering: How can we build a model with what we don’t know?

• Computing with uncertainty: What would a unified uncertainty calculator look like?

• Admitting you don’t know: How should epistemic uncertainty be communicated?

• Manifesto on epistemic uncertainty: Can we agree on anything?

The event features several themed sessions, focused by addresses by international thought leaders followed by open discussion on one of these questions.

You can register for the event by clicking here. Attendance is free to all.

Decision making, 2 February 2021, 14:00-16:30 GMT

Anthony O'Hagan, We Already Have a Unified Uncertainty Theory

William Oberkampf, Simulation-Informed Decision Making

Discussion: Must decisions and designs distinguish kinds of uncertainties?

Computing with uncertainty, 3 February, 16:00-18:30 GMT

Min-ge Xie, Bayesian/Fiducial/Frequentist Uncertainty Quantification by Artificial Samples

Vladik Kreinovich, Epistemic vs Aleatory: Granular Computing and Ideas Beyond That

Discussion: What would a unified uncertainty calculator look like?

Manifesto on uncertainty, 15 February, 10:00-12:30 GMT

Thierry Denoeux, Random Fuzzy Sets: A General Model of Epistemic Uncertainty

Michael Goldstein, The Basic Principles of Reasoning About Uncertainty: A Subjectivist Approach

Discussion: Can we agree on anything about handling epistemic uncertainty in engineering?

Kinds of uncertainty, 17 February, 16:00-18:30 GMT

Yakov Ben-Haim, Innovation Dilemmas: An Info-gap Perspective

Francois Hemez, Uncertainty in Numerical Simulations: Does It Matter? What To Do About It?

Discussion: Are there other kinds of uncertainty?

Uncertainty engineering, 3 March, 15:00-17:30 GMT

Daniel Straub, Epistemic Uncertainty in Engineering Decision Making

Luis Crespo, Uncertainty Modelling and Optimization Under Uncertainty

Discussion: How can we build a model given what we don’t know?

Manifesto on uncertainty (revisited), 12 March, 14:00-16:30 GMT

Terje Aven, Conceptualising, Representing and Describing Epistemic Uncertainties in Risk Analysis

Keith Worden, Scott Ferson, Say What?

Discussion: Can we agree on anything about handling epistemic uncertainty in engineering?

Uncertainty arithmetic, 17 March, 15:00-17:30 GMT

Didier Dubois, Unified View of Uncertainty Theories [tentative]

Laura Swiler, Epistemic Uncertainty: Computation and Usage

Discussion: Do epistemic and aleatory need different calculi?

Admitting what you don't know, 25 March 16:00-18:30 GMT

Gerd Gigerenzer, Risk Literacy and Health

Ann Bostrom, Ways people talk about risk and uncertainty

Discussion: How should we explain and talk about uncertainty?

Management of uncertainty, 20th April, 14:00-16:30 GMT

Eleni Chatzi, Vasilis Dertimanis, Counting on Uncertain Models for Structural Health Monitoring

Michael Beer, Epistemic Uncertainties: Opportunities or Burden?

Discussion: How are management, compliance and monitoring affected by uncertainties?

ZOOM

To participate in the conference live, you will need to download a small, free app from Zoom. You join the meeting by clicking on the link that will be supplied by email acknowledging your registration, or by entering a multidigit code. You can connect over the web if your device has a speaker, or connect audio by telephone if it doesn’t. You’ll need a camera (if you want to be seen) and a microphone (if you want to be heard).

You can even access the conference via your smartphone with internet access using Zoom downloaded from the app store, although detailed presentation slides might be rather small on your phone display. If you want to participate in the discussion, you would have to grant Zoom permission to access your phone’s camera and microphone. You can mute and unmute your microphone any time by moving your cursor to the lower left corner of the screen and clicking on the microphone icon. Turn on or off your camera similarly. There is also a chat facility by which you can pose questions and make silent interjections during discussions or presentations. You can leave a Zoom meeting anytime by clicking on the red button in the lower, right corner of the Zoom screen. You can rejoin the meeting later if it is still going on.

Video recordings of the talks and discussions will be archived at http://riskinstitute.uk/.

DEDICATION

ViCE is dedicated to B. John Garrick, who passed away unexpectedly in 2020 after a long career devoted to bringing the tools of probabilistic analysis to technological problems across engineering. The following is a panegyric commenting on B.J. Garrick’s 2010 paper “Commentary Interval Analysis Versus Probabilistic Analysis” (which appeared in Risk Analysis 30: 369-370. DOI: 10.1111/j.1539-6924.2010.01360.x), which was a critique of Terje Aven’s 2010 perspective paper “On the Need for Restricting the Probabilistic Analysis in Risk Assessments to Variability” (in Risk Analysis 30:354-360. DOI: 10.1111/j.1539-6924.2009.01314.x).

B. John Garrick was one of the founders of modern risk science, the primary goal of which, he thought, is to improve decision making. He held that probability as a thought process should be used to foster honesty in the analysis, “to better tell the truth about threats, events, and their outcome”. Garrick shined a light on the ways in which stochasticity impacts technological advancement. Deterministic analysis is insufficient for an increasing number of complex applications, and a practical consideration of risk analysis can make strides in safety, economy and performance in many disciplines.

Some researchers, like Terje Aven, have argued that a more expressive representation beyond traditional Laplacian probability is needed for problems combining epistemic and aleatory uncertainties. Garrick saw this argument as losing sight of the primary goal, and of improperly omitting valuable details. He noted that the additional benefits of a new theory of imprecise probabilities pale in comparison to those of a straightforward informative probabilistic approach.

Of course accepting the idea that there could be non-Laplacian uncertainty doesn’t mean one throws away probability theory, any more than accepting non-Euclidean geometry removes the need for Euclidean geometry. The traditional approach remains as practically important and useful as before, with just as many applications. Indeed, the probabilistic approach Garrick championed will surely remain the primary tool in risk analysis, fully appropriate for a host of problems. The new imprecise approach relaxes one of the assumptions of the traditional theory, and it is a generalisation of probability. But Garrick’s point, that it seems unnecessary for the main use of risk analysis, is a serious one. Do the nuances or possibly wider applications pay for the complications inherent in generalising probability theory?

The issues at the very heart of the debate exemplified by these papers by Garrick and Aven are the focus of the ViCE virtual conference on epistemic uncertainty in engineering, which has been dedicated to Garrick in appreciation of his contributions to the subject.

SPONSORS