Skip to main content

Max Henrion

Carnegie Mellon University, Engineering and Public Policy, Adjunct

Followers

65

Following

16

Co-authors

12

Public Views

CEO and Founder, Lumina Decision Systems. Adjunct Professor (formerly Associate Professor) at Carnegie Mellon, Department of Engineering and Public Policy. Founding President Association of Uncertainty and Artificial Intelligence. PhD Carnegie Mellon University, MDes Royal College of Art, London, BA Cambridge University.

less

Universidade Estadual de Campinas

Heather Douglas

Michigan State University

University of Kent

Jennifer Gidley

University of Technology Sydney

Kansas State University

University of Regensburg

michele ciavarella

Politecnico di Bari

Armando Marques-Guedes

UNL - New University of Lisbon

María Elena Hernández Ramírez

University of Guadalajara

Raul Pacheco-Vega

Flacso México

InterestsView All (6)

Uploads

Papers by Max Henrion

In GLIDE, we have attempted to combine the most desirable mix of these features that would be use... more

Acquiring knowledge relevant to decision making

Search-based Methods to Bound Diagnostic Probabilities in Very Large Belief Nets

arXiv (Cornell University), Mar 20, 2013

Since exact probabilistic inference is intractable in general for large multiply connected belief... more Since exact probabilistic inference is intractable in general for large multiply connected belief nets, approximate methods are required. A promising approach is to use heuristic search among hypotheses (instantiations of the network) to find the most probable ones, as in the TopN algorithm. Search is based on the relative probabilities of hypotheses which are efficient to compute. Given upper and lower bounds on the relative probability of partial hypotheses, it is possible to obtain bounds on the absolute probabilities of hypotheses. Best-first search aimed at reducing the maximum error progressively narrows the bounds as more hypotheses are examined. Here, qualitative probabilistic analysis is employed to obtain bounds on the relative probability of partial hypotheses for the BN20 class of networks networks and a generalization replacing the noisy OR assumption by negative synergy. The approach is illustrated by application to a very large belief network, QMR-BN, which is a reformulation of the Internist-! system for diagnosis in internal medicine.

Assessing Uncertainty in Physical Constants

Cambridge University Press eBooks, Jul 8, 2002

Methods for R&D Portfolio Analysis and Evaluation (Workshop Report)

This report is available at no cost from the National Renewable Energy Laboratory (NREL) at www.n... more

How Much More Probable is "Much More Probable"? Verbal Expressions for Probability Updates

arXiv (Cornell University), Mar 27, 2013

Abstr"ct-Bayesian inference systems should be able to explain their reasoning to users, translati... more Abstr"ct-Bayesian inference systems should be able to explain their reasoning to users, translating from numerical to natural language. Previous empirical work has investigated the correspondence between absolute probabilities and linguistic phrases. This study extends that work to the correspondence between changes in probabilities (u pdates) and relative probability phrases, such as ''much more likely" or "a little less likely." Subjects selected such phrases to best describe numerical probability updates. We examined three hypotheses about the correspondence, and found the most descriptively accurate of these three to be that each such phrase corresponds to a fixed difference in probability (rather tha n fixed ratio of probabilities or of odds). The empirically derived phrase selection function uses eight phrases and achieved a 72% accuracy in correspondence with the subjects' actual usage.

Efficient Estimation of the Value of Information in Monte Carlo Models

The expected value of information (EVI) is the most powerful measure of sensitivity to uncer tain... more The expected value of information (EVI) is the most powerful measure of sensitivity to uncer tainty in a decision model: it measures the potential of information to improve the decision, and hence measures the expected value of the outcome. Stan dard methods for computing EVI use discrete vari ables and are computationally intractable for models that contain more than a few variables. Monte Carlo simulation provides the basis for more tractable evaluation of large predictive models with continu ous and discrete variables, but so far computation of EVI in a Monte Carlo setting also has appeared impractical. We introduce an approximate approach based on preposterior analysis for estimating EVI in Monte Carlo models. Our method uses a linear approximation to the value function and multiple linear regression to estimate the linear model from the samples. The approach is efficient and practical for extremely large models. It allows easy estima tion of EVI for perfect or panial information on individual variables or on combinations of variables. We illustrate its implementation within Demos (a decision modeling system), and its application to a large model for crisis transponation planning.

Practical issues in constructing a Bayes belief network

International Journal of Approximate Reasoning, Jul 1, 1988

Bayes belief networks and influence diagrams are tools for constructing coherent probabilistic re... more Bayes belief networks and influence diagrams are tools for constructing coherent probabilistic representations of uncertain expert opinion. The construction of such a network with about 30 nodes is used to illustrate a variety of techniques which can facilitate the process of structuring and quantifying uncertain relationships. These include some generalizations of the "noisy OR gate" concept. Sensitivity analysis of generic elements of Bayes' networks provides insight into when rough probability assessments are sufficient and when greater precision may be important. 1. Introduction

Probabilistic Diagnosis Using a Reformulation of the INTERNIST-1/QMR Knowledge Base

Methods of Information in Medicine, 1991

The value of knowing how little you know : the advantages of a probabilistictreatment of uncerainty in policy analysis

University Microfilms eBooks, 1982

An Empirical Study of Strategies for Understanding Quantitative Decision Models

Demos is a modeling environment designed to help a co-operating team design, analyze, critique an... more Demos is a modeling environment designed to help a co-operating team design, analyze, critique and refine quantitative models for policy research. Earlier research found that readers of Demos models tended to become disoriented while exploring models online. In response we have designed and implemented a graphical interface to Demos named Demaps. Demaps displays diagrams of the model structure, both dependence networks and abstraction hierarchies, to provide graphic context and direct manipulation style of interaction. We describe a study of the use of Demaps to understand and compare multiple versions of models. The study employs verbal protocol analysis to evaluate the design of Demaps and to discover expert strategies for model understanding and criticism. Subjects were able to learn to use Demaps effectively in about an hour to review and compare policy models and perform sensitivity analyses. The study describes two strategies used in reading models and suggests the desirabilit...

Proceedings of the Fourth Annual Conference on Uncertainty in Artificial Intelligence

Decision Support Models to Integrate Market and Non-Market Value Attributes for Platform Decommissioning: An Effective Approach for Resolving Challenges at the Nexus of Science and Regulatory Policymaking

Energy sector portfolio analysis with uncertainty

Applied Energy, 2022

Authors group, 2019

Integrating Decision Support Models with Market and Non-Market Value Attributes for Platform Decommissioning: An Effective Approach for Resolving the Challenges Inherent at the Nexus of Science and Policy

Offshore Technology Conference, 2017

Many critical factors must be considered in determining the most feasible and appropriate options... more Many critical factors must be considered in determining the most feasible and appropriate options for decommissioning offshore oil and gas platforms, including alternatives from complete removal to partial removal with artificial reefing. Using financial-based cost/benefit analysis to evaluate decommissioning options neglects the complexities inherent in a successful platform decommissioning process, which includes non-monetary values such as the production of greenhouse gasses and the gains and/or losses of ecological services due to partial or complete removal. Multi-attribute decision analysis provides technically defensible consideration of both market and non-market values associated with the outcomes of rig decommissioning while allowing stakeholders to clearly and transparently investigate and evaluate the nexus between financial considerations and other factors (e.g., air/water quality, biological productivity, ocean access) that drive public policy. In California, the PLATF...

A Decision Support Model for Platform Decommissioning: Successful Applications in California and Implications for Worldwide Use

SPE International Conference and Exhibition on Health, Safety, Security, Environment, and Social Responsibility, 2016

As companies around the world plan for the future decommissioning of oil and gas platforms, a cas... more As companies around the world plan for the future decommissioning of oil and gas platforms, a case study from California provides a model in terms of both process and the use of decision support tools. The question of how to decommission California's 27 offshore oil platforms, like many issues related to offshore oil and gas in California, began as a major public controversy. We created a decision analysis software tool (PLATFORM) to clarify and evaluate decision strategies as part of a larger policy analysis conducted for the California Ocean Science Trust. The in-depth analysis focused on two main options—complete platform removal and partial removal to 85 feet below the water line, with the remaining structures to be converted in place as an artificial reef to preserve their biological value. PLATFORM was key in structuring and performing analyses of the impacts of each option (e.g., on costs, fishery production, air emissions) and greatly improved the team's productivity...

Atmospheric Deposition Assessment Model: Applications to Regional Aquatic Acidification in Eastern North America

Impact Models to Assess Regional Acidification, 1990

Low-Carbon Development

Some rights reserved 1 2 3 4 16 15 14 13 This work is a product of the staff of The World Bank wi... more Some rights reserved 1 2 3 4 16 15 14 13 This work is a product of the staff of The World Bank with external contributions. Note that The World Bank does not necessarily own each component of the content included in the work. The World Bank therefore does not warrant that the use of the content contained in the work will not infringe on the rights of third parties. The risk of claims resulting from such infringement rests solely with you. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved.

Technometrics, 1995

In GLIDE, we have attempted to combine the most desirable mix of these features that would be use... more

Acquiring knowledge relevant to decision making

Search-based Methods to Bound Diagnostic Probabilities in Very Large Belief Nets

arXiv (Cornell University), Mar 20, 2013

Since exact probabilistic inference is intractable in general for large multiply connected belief... more Since exact probabilistic inference is intractable in general for large multiply connected belief nets, approximate methods are required. A promising approach is to use heuristic search among hypotheses (instantiations of the network) to find the most probable ones, as in the TopN algorithm. Search is based on the relative probabilities of hypotheses which are efficient to compute. Given upper and lower bounds on the relative probability of partial hypotheses, it is possible to obtain bounds on the absolute probabilities of hypotheses. Best-first search aimed at reducing the maximum error progressively narrows the bounds as more hypotheses are examined. Here, qualitative probabilistic analysis is employed to obtain bounds on the relative probability of partial hypotheses for the BN20 class of networks networks and a generalization replacing the noisy OR assumption by negative synergy. The approach is illustrated by application to a very large belief network, QMR-BN, which is a reformulation of the Internist-! system for diagnosis in internal medicine.

Assessing Uncertainty in Physical Constants

Cambridge University Press eBooks, Jul 8, 2002

Methods for R&D Portfolio Analysis and Evaluation (Workshop Report)

This report is available at no cost from the National Renewable Energy Laboratory (NREL) at www.n... more

How Much More Probable is "Much More Probable"? Verbal Expressions for Probability Updates

arXiv (Cornell University), Mar 27, 2013

Abstr"ct-Bayesian inference systems should be able to explain their reasoning to users, translati... more Abstr"ct-Bayesian inference systems should be able to explain their reasoning to users, translating from numerical to natural language. Previous empirical work has investigated the correspondence between absolute probabilities and linguistic phrases. This study extends that work to the correspondence between changes in probabilities (u pdates) and relative probability phrases, such as ''much more likely" or "a little less likely." Subjects selected such phrases to best describe numerical probability updates. We examined three hypotheses about the correspondence, and found the most descriptively accurate of these three to be that each such phrase corresponds to a fixed difference in probability (rather tha n fixed ratio of probabilities or of odds). The empirically derived phrase selection function uses eight phrases and achieved a 72% accuracy in correspondence with the subjects' actual usage.

Efficient Estimation of the Value of Information in Monte Carlo Models

The expected value of information (EVI) is the most powerful measure of sensitivity to uncer tain... more The expected value of information (EVI) is the most powerful measure of sensitivity to uncer tainty in a decision model: it measures the potential of information to improve the decision, and hence measures the expected value of the outcome. Stan dard methods for computing EVI use discrete vari ables and are computationally intractable for models that contain more than a few variables. Monte Carlo simulation provides the basis for more tractable evaluation of large predictive models with continu ous and discrete variables, but so far computation of EVI in a Monte Carlo setting also has appeared impractical. We introduce an approximate approach based on preposterior analysis for estimating EVI in Monte Carlo models. Our method uses a linear approximation to the value function and multiple linear regression to estimate the linear model from the samples. The approach is efficient and practical for extremely large models. It allows easy estima tion of EVI for perfect or panial information on individual variables or on combinations of variables. We illustrate its implementation within Demos (a decision modeling system), and its application to a large model for crisis transponation planning.

Practical issues in constructing a Bayes belief network

International Journal of Approximate Reasoning, Jul 1, 1988

Bayes belief networks and influence diagrams are tools for constructing coherent probabilistic re... more Bayes belief networks and influence diagrams are tools for constructing coherent probabilistic representations of uncertain expert opinion. The construction of such a network with about 30 nodes is used to illustrate a variety of techniques which can facilitate the process of structuring and quantifying uncertain relationships. These include some generalizations of the "noisy OR gate" concept. Sensitivity analysis of generic elements of Bayes' networks provides insight into when rough probability assessments are sufficient and when greater precision may be important. 1. Introduction

Probabilistic Diagnosis Using a Reformulation of the INTERNIST-1/QMR Knowledge Base

Methods of Information in Medicine, 1991

The value of knowing how little you know : the advantages of a probabilistictreatment of uncerainty in policy analysis

University Microfilms eBooks, 1982

An Empirical Study of Strategies for Understanding Quantitative Decision Models

Demos is a modeling environment designed to help a co-operating team design, analyze, critique an... more Demos is a modeling environment designed to help a co-operating team design, analyze, critique and refine quantitative models for policy research. Earlier research found that readers of Demos models tended to become disoriented while exploring models online. In response we have designed and implemented a graphical interface to Demos named Demaps. Demaps displays diagrams of the model structure, both dependence networks and abstraction hierarchies, to provide graphic context and direct manipulation style of interaction. We describe a study of the use of Demaps to understand and compare multiple versions of models. The study employs verbal protocol analysis to evaluate the design of Demaps and to discover expert strategies for model understanding and criticism. Subjects were able to learn to use Demaps effectively in about an hour to review and compare policy models and perform sensitivity analyses. The study describes two strategies used in reading models and suggests the desirabilit...

Proceedings of the Fourth Annual Conference on Uncertainty in Artificial Intelligence

Decision Support Models to Integrate Market and Non-Market Value Attributes for Platform Decommissioning: An Effective Approach for Resolving Challenges at the Nexus of Science and Regulatory Policymaking

Energy sector portfolio analysis with uncertainty

Applied Energy, 2022

Authors group, 2019

Integrating Decision Support Models with Market and Non-Market Value Attributes for Platform Decommissioning: An Effective Approach for Resolving the Challenges Inherent at the Nexus of Science and Policy

Offshore Technology Conference, 2017

Many critical factors must be considered in determining the most feasible and appropriate options... more Many critical factors must be considered in determining the most feasible and appropriate options for decommissioning offshore oil and gas platforms, including alternatives from complete removal to partial removal with artificial reefing. Using financial-based cost/benefit analysis to evaluate decommissioning options neglects the complexities inherent in a successful platform decommissioning process, which includes non-monetary values such as the production of greenhouse gasses and the gains and/or losses of ecological services due to partial or complete removal. Multi-attribute decision analysis provides technically defensible consideration of both market and non-market values associated with the outcomes of rig decommissioning while allowing stakeholders to clearly and transparently investigate and evaluate the nexus between financial considerations and other factors (e.g., air/water quality, biological productivity, ocean access) that drive public policy. In California, the PLATF...

A Decision Support Model for Platform Decommissioning: Successful Applications in California and Implications for Worldwide Use

SPE International Conference and Exhibition on Health, Safety, Security, Environment, and Social Responsibility, 2016

As companies around the world plan for the future decommissioning of oil and gas platforms, a cas... more As companies around the world plan for the future decommissioning of oil and gas platforms, a case study from California provides a model in terms of both process and the use of decision support tools. The question of how to decommission California's 27 offshore oil platforms, like many issues related to offshore oil and gas in California, began as a major public controversy. We created a decision analysis software tool (PLATFORM) to clarify and evaluate decision strategies as part of a larger policy analysis conducted for the California Ocean Science Trust. The in-depth analysis focused on two main options—complete platform removal and partial removal to 85 feet below the water line, with the remaining structures to be converted in place as an artificial reef to preserve their biological value. PLATFORM was key in structuring and performing analyses of the impacts of each option (e.g., on costs, fishery production, air emissions) and greatly improved the team's productivity...

Atmospheric Deposition Assessment Model: Applications to Regional Aquatic Acidification in Eastern North America

Impact Models to Assess Regional Acidification, 1990

Low-Carbon Development

Some rights reserved 1 2 3 4 16 15 14 13 This work is a product of the staff of The World Bank wi... more Some rights reserved 1 2 3 4 16 15 14 13 This work is a product of the staff of The World Bank with external contributions. Note that The World Bank does not necessarily own each component of the content included in the work. The World Bank therefore does not warrant that the use of the content contained in the work will not infringe on the rights of third parties. The risk of claims resulting from such infringement rests solely with you. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved.

Technometrics, 1995