Linda C. Van Der Gaag - Profile on Academia.edu (original) (raw)
Papers by Linda C. Van Der Gaag
International Journal of Approximate Reasoning, Nov 1, 2020
Immunohistochemistry (IHC) was performed on tissue samples of untreated (n¼40) and chemotherapypr... more Immunohistochemistry (IHC) was performed on tissue samples of untreated (n¼40) and chemotherapypretreated (n¼14) MPM patients. Different subsets if immune cells were identified based on staining for CD4, CD8, FoxP3, CD68, CD45RO and granzyme B. The expression of the immune checkpoints TIM-3, LAG-3, PD-1 and its ligand PD-L1 was also investigated. The relationship between the immunological parameters and survival, as well as response to chemotherapy was analyzed using the R statistical software. Results: All patients had CD8+ tumor infiltrating lymphocytes (TILs), CD68+ histiocytes and macrophages and CD45RO+ T cells in their stroma, with CD8+ TILs being the predominant cell type of the immune infiltrate. Stromal CD4+ TILs were found in 75% of the untreated and 71% of the pretreated samples. A subset of those cells was also FoxP3+ and these CD4+FoxP3+ cells were positively correlated with stromal CD4 expression (p<0.001). Less than half of the samples showed positivity for granzyme B. Both, untreated and pretreated patients had PD-1+ TILs, while only 10% of the untreated patients also had PD-1+ tumor cells. PD-L1 positivity on lymphocytes and/or tumor cells was observed for more than half of the patients, with significant differences according to the histological subtype (p<0.001). Patients with a sarcomatoid histology showed the most PD-1 expression. TIM-3 was expressed in tumor cells, stromal lymphocytes and plasma cells, less often in pretreated samples compared to untreated samples. All samples were negative for LAG-3. After multivariate analysis stromal CD45RO expression was found to be an independent negative predictive factor for response to chemotherapy (p¼0.017) and expression of CD4 and TIM-3 in lymphoid aggregates were good prognostic factors (p¼0.008; p¼0.001). Our data reveal the diversity of immune cells present in MPM and point to TIM-3 as a new target in mesothelioma. Administering chemotherapy before or together with PD-1/PD-L1/TIM-3 blocking agents may not be the best combination sequence and further research on the predictive value of CD45RO in the stroma might guide patient selection for chemotherapy.
Department of Computer Science [CS], 1986
For several years, Lisp has been the most popular programming language for artificial intelligenc... more For several years, Lisp has been the most popular programming language for artificial intelligence. PRO-LOG, however, is rapidly becoming the second most popular Al language; for several applications, PRO-LOG is even preferred to Lisp. In this paper, the suitability of PROLOG as an expert system building tool is demonstrated: a small expert system shell is discussed, and compared to the OELFl-2 system, after the example of which the PROLOG system has been developed.
Generalized Rules of Probabilistic Independence
Lecture Notes in Computer Science, 2021
UU WINFI Informatica en Informatiekunde eBooks, 2005
The process of engineering probabilistic networks can be supported by a library of generic knowle... more The process of engineering probabilistic networks can be supported by a library of generic knowledge structures. Such a knowledge structure is instantiated with domainspecific knowledge and is used to derive, in a number of steps, a segment of the graphical structure of a network. To provide for customisation to the application at hand, the structures are based on an in-depth knowledge analysis and capture, in an appropriate representation, the intricate details of the knowledge involved. We present, as an example, the generic knowledge structure that captures the relations between a test result and the underlying true value. As a guideline for its application we provide the derivation of a network segment in the field of oncology.
Utrecht University: Information and Computing Sciences eBooks, 1996
Uncertainty in Artificial Intelligence, Jul 27, 1990
Many AI researchers argue that probability theory is only capable of dealing with uncertainty in ... more Many AI researchers argue that probability theory is only capable of dealing with uncertainty in situations where a fully specified joint probability distribution is available, and conclude that it is not suitable for application in AI systems. Probability intervals, however, constitute a means for expressing incompleteness of information. We present a method for computing probability interval! for probabilities of interest from a partial specification of a joint probability distribution. Our method improves on earlier approaches by all owing for independency relation ships between statistical variables to be exploited .
Using com-specific risks to support the detection of clinical mastitis on farms with an atomatic milking system
Department of Computer Science [CS], 1988
Symbolic and Quantiative Approaches to Resoning with Uncertainty: 12th European Conference, ECSQARU 2013, Utrecht, The Netherlands, July 8-10, 2013, Proceedings
Springer eBooks, Jun 24, 2013
A Formal Concept View of Abstract Argumentation.- Approximating Credal Network Inferences by Line... more A Formal Concept View of Abstract Argumentation.- Approximating Credal Network Inferences by Linear Programming.- A Comparative Study of Compilation-Based Inference Methods for Min-Based Possibilistic Networks.- Qualitative Combination of Independence Models.- A Case Study on the Application of Probabilistic Conditional Modelling and Reasoning to Clinical Patient Data in Neurosurgery.- Causal Belief Networks: Handling Uncertain Interventions.- On Semantics of Inference in Bayesian Networks.- Evaluating Asymmetric Decision Problems with Binary Constraint Trees.- On the Equivalence between Logic Programming Semantics and Argumentation Semantics.- A Fuzzy-Rough Data Pre-processing Approach for the Dendritic Cell Classifier.- Compiling Probabilistic Graphical Models Using Sentential Decision Diagrams.- Independence in Possibility Theory under Different Triangular Norms.- Probabilistic Satisfiability and Coherence Checking through Integer Programming.- Extreme Lower Previsions and Minkowski Indecomposability.- Qualitative Capacities as Imprecise Possibilities.- Conditional Preference Nets and Possibilistic Logic.- Many-Valued Modal Logic and Regular Equivalences in Weighted Social Networks.- Zero-Probability and Coherent Betting: A Logical Point of View.- Conditional Random Quantities and Iterated Conditioning in the Setting of Coherence.- Distance-Based Measures of Inconsistency.- Safe Probability: Restricted Conditioning and Extended Marginalization.- Maximin Safety: When Failing to Lose Is Preferable to Trying to Win.- Weighted Regret-Based Likelihood: A New Approach to Describing Uncertainty.- Structural Properties for Deductive Argument Systems.- Measuring Inconsistency through Minimal Proofs.- Representing Synergy among Arguments with Choquet Integral.- A Reasoning Platform Based on the MI Shapley Inconsistency Value.- Most Inforbable Explanations: Finding Explanations in Bayesian Networks That Are Both Probable and Informative.- Structure Approximation of Most Probable Explanations in Bayesian Networks.- Argumentation Based Dynamic Multiple Criteria Decision Making.- Conditional Beliefs in a Bipolar Framework.- Detecting Marginal and Conditional Independencies between Events and Learning Their Causal Structure.- Measuring Incompleteness under Multi-valued Semantics by Partial MaxSAT Solvers.- On the Tree Structure Used by Lazy Propagation for Inference in Bayesian Networks.- Hierarchical Model for Rank Discrimination Measures.- Extreme Points of the Credal Sets Generated by Elementary Comparative Probabilities.- MCMC Estimation of Conditional Probabilities in Probabilistic Programming Languages.- Sorted-Pareto Dominance and Qualitative Notions of Optimality.- A First-Order Dynamic Probability Logic.- Selecting Source Behavior in Information Fusion on the Basis of Consistency and Specificity.- On the Problem of Reversing Relational Inductive Knowledge Representation.- Analogical Proportions and Multiple-Valued Logics.- Chain Graph Interpretations and Their Relations.- On the Plausibility of Abstract Arguments.
International Journal of Intelligent Systems, 1998
The concept of informational independence plays a key role in most knowledge-based systems. J. Pe... more The concept of informational independence plays a key role in most knowledge-based systems. J. Pearl and his co-researchers have analysed the basic properties of the concept and have formulated an axiomatic system for informational independence. This axiomatic system focuses on independences among mutually disjoint sets of variables. We show that in the context of probabilistic independence a focus on disjoint sets of variables can hide various interesting properties. To capture these properties, we enhance Pearl's axiomatic system with two additional axioms. We investigate the set of models of the thus enhanced system and show that it provides a better characterisation of the concept of probabilistic independence than Pearl's system does. In addition, we observe that both Pearl's axiomatic system and our enhanced system o er inference rules for deriving new independences from an initial set of independence statements and as such allow for a normal form for representing independence. We address the normal forms ensuing from the two axiomatic systems for informational independence.
Uncertainty in artificial intelligence : proceedings of the Twenty-third Conference (2007) : July 19-22, 2007, Vancouver, British Columbia
AUAI Press eBooks, 2007
Department of Computer Science [CS], 1989
The Centre for Mathematics and Computer Science is a research institute of the Stichting Mathemat... more The Centre for Mathematics and Computer Science is a research institute of the Stichting Mathematisch Centrum, which was founded on February 11, 1946, as a nonprofit institution aiming at the promotion of mathematics, computer science, and their applications. It is sponsored by the Dutch Govern- ment through the Netherlands Organization for the Advancement of Research (N.W.O.).
Evidence absorption : experiments on different classes of randomly generated belief networks
Unknown Publisher eBooks, 1994
ABSTRACT
Department of Computer Science [CS], 1987
Most expert knowledge is of an ill-defined and heuristic nature. Therefore, many present-day rule... more Most expert knowledge is of an ill-defined and heuristic nature. Therefore, many present-day rule-based expert systems include a mechanism for modelling and manipulating imprecise knowledge. For a long time, probability theory has been the primary quantitative approach for handling uncertainty. Other mathematical models of uncertainty have been proposed during the last decade, several of which depart from probability theory. The certainty factor model proposed by the authors of the MYCIN system is an example of an ad hoe model. The aim in developing the model was primarily to develop a method that was of practical use. The certainty factor model is computationally simple, a property that has led to its considerable success. In this paper, we use so-called inference networks to demonstrate the application of the model in a rule-based top-down reasoning expert system. This approach enables us to show some inadequacies of the notational convention used by the creators of the model. We propose a syntactically correct formalism and use this formalism to discuss several properties of the model.
Utrecht University: Information and Computing Sciences eBooks, 1996
Probabilistic inference with a belief network in general is computationally expensive. Since the ... more Probabilistic inference with a belief network in general is computationally expensive. Since the concept of structural relevance provides for identifying parts of a belief network that are irrelevant to a context of interest, it allows for alleviating to some extent the computational burden of inference: inference can be restricted to the network's relevant part. The structurally relevant part of a belief network, however, is not static. It may c hange dynamically as reasoning progresses. We address the dynamics of structural relevance and introduce the concept of an independence projection to capture these dynamics.
Congress of the Italian Association for Artificial Intelligence, Sep 14, 1999
arXiv (Cornell University), Jun 13, 2012
While known algorithms for sensitivity analysis and parameter tuning in probabilistic networks ha... more While known algorithms for sensitivity analysis and parameter tuning in probabilistic networks have a running time that is exponential in the size of the network, the exact computational complexity of these problems has not been established as yet. In this paper we study several variants of the tuning problem and show that these problems are NPPP-complete in general. We further show that the problems remain NP-complete or PP-complete, for a number of restricted variants. These complexity results provide insight in whether or not recent achievements in sensitivity analysis and tuning can be extended to more general, practicable methods.
Proceedings of the 12th European conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty
Invited Papers.- Graphical Models as Languages for Computer Assisted Diagnosis and Decision Makin... more Invited Papers.- Graphical Models as Languages for Computer Assisted Diagnosis and Decision Making.- Planning with Uncertainty and Incomplete Information.- What&amp;amp;amp;amp;#39;s Your Preference? And How to Express and Implement It in Logic Programming!.- Contributed Papers.- On Preference Representation on an Ordinal Scale.- Rule-Based Decision Support in Multicriteria Choice and Ranking.- Propositional Distances and Preference Representation.- Contributed Papers.- Value Iteration over Belief Subspace.- Space-Progressive Value Iteration: An Anytime Algorithm for a Class of POMDPs.- Contributed Papers.- Reasoning about Intentions in Uncertain Domains.- Troubleshooting with Simultaneous Models.- A Rational Conditional Utility Model in a Coherent Framework.- Contributed Papers.- Probabilistic Reasoning as a General Unifying Tool.- An Operational View of Coherent Conditional Previsions.- Contributed Papers.- Decomposition of Influence Diagrams.- Mixtures of Truncated Exponentials in Hybrid Bayesian Networks.- Importance Sampling in Bayesian Networks Using Antithetic Variables.- Using Recursive Decomposition to Construct Elimination Orders, Jointrees, and Dtrees.- Caveats For Causal Reasoning With Equilibrium Models.- Contributed Papers.- Supporting Changes in Structure in Causal Model Construction.- The Search of Causal Orderings: A Short Cut for Learning Belief Networks.- Stochastic Local Algorithms for Learning Belief Networks: Searching in the Space of the Orderings.- An Empirical Investigation of the K2 Metric.- Contributed Papers.- Sequential Valuation Networks: A New Graphical Technique for Asymmetric Decision Problems.- A Two-Steps Algorithm for Min-Based Possibilistic Causal Networks.- Contributed Papers.- Computing Intervals of Probabilities with Simulated Annealing and Probability Trees.- Probabilistic Logic under Coherence, Model-Theoretic Probabilistic Logic, and Default Reasoning.- Contributed Papers.- Belief Functions with Partially Ordered Values.- Dempster Specialization Matrices and the Combination of Belief Functions.- On the Conceptual Status of Belief Functions with Respect to Coherent Lower Probabilities.- About Conditional Belief Function Independence.- The Evaluation of Sensors&amp;amp;amp;amp;#39; Reliability and Their Tuning for Multisensor Data Fusion within the Transferable Belief Model.- Coarsening Approximations of Belief Functions.- Contributed Papers.- Label Semantics: A Formal Framework for Modeling with Words.- Reasoning about Knowledge Using Rough Sets.- Contributed Papers.- The Capacity of a Possibilistic Channel.- New Semantics for Quantitative Possibility Theory.- Bridging logical, comparative and graphical possibilistic representation frameworks.- Contributed Papers.- Ellipse fitting with uncertainty and fuzzy decision stage for detection. Application in videomicroscopy..- Probabilistic Modelling for Software Quality Control.- Spatial Information Revision: A Comparison between 3 Approaches.- Contributed Papers.- Social Choice, Merging, and Elections.- Data merging: Theory of Evidence vs knowledge-bases merging operators.- Contributed Papers.- A Priori Revision.- Some Operators for Iterated Revision.- On Computing Solutions to Belief Change Scenarios.- &amp;amp;amp;amp;quot;Not impossible&amp;amp;amp;amp;quot; vs. &amp;amp;amp;amp;quot;guaranteed possible&amp;amp;amp;amp;quot; in fusion and revision.- General Preferential Entaulments as Circumscriptions.- Contributed Papers.- A Semantic Tableau Version of First-Order Quasi-Classical Logic.- On Anytime Coherence-Based Reasoning.- Resolving Conflicts between Beliefs, Obligations, Intentions, and Desires.- Contributed Papers.- Comparing a Pair-wise Compatibility Heuristic and Relaxed Stratification: Some Preliminary Results.- How to Reason Credulously and Skeptically within a Single Extension.- Contributed Papers.- Handling Conditionals Adequately in Uncertain Reasoning.- Rankings We Prefer.- Contributed Papers.- Formalizing Human Uncertain Reasoning with Default Rules: A Psychological Conundrum and a Pragmatic Suggestion.- Statistical Information, Uncertainty, and Bayes&amp;amp;amp;amp;#39; Theorem: Some Applications in Experimental Psychology.- Polymorphism of Human Judgment under Uncertainty.- How to Doubt about a Conditional.- Contributed Papers.- Dialectical Proof Theories for the Credulous Preferred Semantics of Argumentation Frameworks.- Argumentation and Qualitative Probabilistic Reasoning Using the Kappa Calculus.- Contributed Papers.- Importance Measures from Reliability Theory for Probabilistic Assumption-Based Reasoning.- Ramification in the Normative Method of Causality.- Simultaneous Events: Conflicts and Preferences.- Orthogonal Relations for Reasoning about Abstract Events.- Contributed Papers.- Explanatory Relations Based on Mathematical Morphology.- Monotonic and Residuated Logic Programs.- Contributed Papers.- A Proof Procedure for Possibilistic Logic Programming with Fuzzy…
PROMUNDI: Probabilistic Multi-knowledge Networks for Diagnosis
International Symposium on Imprecise Probabilities and Their Applications, 2019
The Bayesian Dirichlet equivalent uniform (BDeu) function is a popular score to evaluate the good... more The Bayesian Dirichlet equivalent uniform (BDeu) function is a popular score to evaluate the goodness of a Bayesian network structure given complete categorical data. Despite its interesting properties, such as likelihood equivalence, it does require a prior expressed via a user-defined parameter known as Equivalent Sample Size (ESS), which significantly affects the final structure. We study conditions to obtain prior independence in BDeu-based structure learning. We show in experiments that the amount of data needed to render the learning robust to different ESS values is prohibitively large, even in big data times.
International Journal of Approximate Reasoning, Nov 1, 2020
Immunohistochemistry (IHC) was performed on tissue samples of untreated (n¼40) and chemotherapypr... more Immunohistochemistry (IHC) was performed on tissue samples of untreated (n¼40) and chemotherapypretreated (n¼14) MPM patients. Different subsets if immune cells were identified based on staining for CD4, CD8, FoxP3, CD68, CD45RO and granzyme B. The expression of the immune checkpoints TIM-3, LAG-3, PD-1 and its ligand PD-L1 was also investigated. The relationship between the immunological parameters and survival, as well as response to chemotherapy was analyzed using the R statistical software. Results: All patients had CD8+ tumor infiltrating lymphocytes (TILs), CD68+ histiocytes and macrophages and CD45RO+ T cells in their stroma, with CD8+ TILs being the predominant cell type of the immune infiltrate. Stromal CD4+ TILs were found in 75% of the untreated and 71% of the pretreated samples. A subset of those cells was also FoxP3+ and these CD4+FoxP3+ cells were positively correlated with stromal CD4 expression (p<0.001). Less than half of the samples showed positivity for granzyme B. Both, untreated and pretreated patients had PD-1+ TILs, while only 10% of the untreated patients also had PD-1+ tumor cells. PD-L1 positivity on lymphocytes and/or tumor cells was observed for more than half of the patients, with significant differences according to the histological subtype (p<0.001). Patients with a sarcomatoid histology showed the most PD-1 expression. TIM-3 was expressed in tumor cells, stromal lymphocytes and plasma cells, less often in pretreated samples compared to untreated samples. All samples were negative for LAG-3. After multivariate analysis stromal CD45RO expression was found to be an independent negative predictive factor for response to chemotherapy (p¼0.017) and expression of CD4 and TIM-3 in lymphoid aggregates were good prognostic factors (p¼0.008; p¼0.001). Our data reveal the diversity of immune cells present in MPM and point to TIM-3 as a new target in mesothelioma. Administering chemotherapy before or together with PD-1/PD-L1/TIM-3 blocking agents may not be the best combination sequence and further research on the predictive value of CD45RO in the stroma might guide patient selection for chemotherapy.
Department of Computer Science [CS], 1986
For several years, Lisp has been the most popular programming language for artificial intelligenc... more For several years, Lisp has been the most popular programming language for artificial intelligence. PRO-LOG, however, is rapidly becoming the second most popular Al language; for several applications, PRO-LOG is even preferred to Lisp. In this paper, the suitability of PROLOG as an expert system building tool is demonstrated: a small expert system shell is discussed, and compared to the OELFl-2 system, after the example of which the PROLOG system has been developed.
Generalized Rules of Probabilistic Independence
Lecture Notes in Computer Science, 2021
UU WINFI Informatica en Informatiekunde eBooks, 2005
The process of engineering probabilistic networks can be supported by a library of generic knowle... more The process of engineering probabilistic networks can be supported by a library of generic knowledge structures. Such a knowledge structure is instantiated with domainspecific knowledge and is used to derive, in a number of steps, a segment of the graphical structure of a network. To provide for customisation to the application at hand, the structures are based on an in-depth knowledge analysis and capture, in an appropriate representation, the intricate details of the knowledge involved. We present, as an example, the generic knowledge structure that captures the relations between a test result and the underlying true value. As a guideline for its application we provide the derivation of a network segment in the field of oncology.
Utrecht University: Information and Computing Sciences eBooks, 1996
Uncertainty in Artificial Intelligence, Jul 27, 1990
Many AI researchers argue that probability theory is only capable of dealing with uncertainty in ... more Many AI researchers argue that probability theory is only capable of dealing with uncertainty in situations where a fully specified joint probability distribution is available, and conclude that it is not suitable for application in AI systems. Probability intervals, however, constitute a means for expressing incompleteness of information. We present a method for computing probability interval! for probabilities of interest from a partial specification of a joint probability distribution. Our method improves on earlier approaches by all owing for independency relation ships between statistical variables to be exploited .
Using com-specific risks to support the detection of clinical mastitis on farms with an atomatic milking system
Department of Computer Science [CS], 1988
Symbolic and Quantiative Approaches to Resoning with Uncertainty: 12th European Conference, ECSQARU 2013, Utrecht, The Netherlands, July 8-10, 2013, Proceedings
Springer eBooks, Jun 24, 2013
A Formal Concept View of Abstract Argumentation.- Approximating Credal Network Inferences by Line... more A Formal Concept View of Abstract Argumentation.- Approximating Credal Network Inferences by Linear Programming.- A Comparative Study of Compilation-Based Inference Methods for Min-Based Possibilistic Networks.- Qualitative Combination of Independence Models.- A Case Study on the Application of Probabilistic Conditional Modelling and Reasoning to Clinical Patient Data in Neurosurgery.- Causal Belief Networks: Handling Uncertain Interventions.- On Semantics of Inference in Bayesian Networks.- Evaluating Asymmetric Decision Problems with Binary Constraint Trees.- On the Equivalence between Logic Programming Semantics and Argumentation Semantics.- A Fuzzy-Rough Data Pre-processing Approach for the Dendritic Cell Classifier.- Compiling Probabilistic Graphical Models Using Sentential Decision Diagrams.- Independence in Possibility Theory under Different Triangular Norms.- Probabilistic Satisfiability and Coherence Checking through Integer Programming.- Extreme Lower Previsions and Minkowski Indecomposability.- Qualitative Capacities as Imprecise Possibilities.- Conditional Preference Nets and Possibilistic Logic.- Many-Valued Modal Logic and Regular Equivalences in Weighted Social Networks.- Zero-Probability and Coherent Betting: A Logical Point of View.- Conditional Random Quantities and Iterated Conditioning in the Setting of Coherence.- Distance-Based Measures of Inconsistency.- Safe Probability: Restricted Conditioning and Extended Marginalization.- Maximin Safety: When Failing to Lose Is Preferable to Trying to Win.- Weighted Regret-Based Likelihood: A New Approach to Describing Uncertainty.- Structural Properties for Deductive Argument Systems.- Measuring Inconsistency through Minimal Proofs.- Representing Synergy among Arguments with Choquet Integral.- A Reasoning Platform Based on the MI Shapley Inconsistency Value.- Most Inforbable Explanations: Finding Explanations in Bayesian Networks That Are Both Probable and Informative.- Structure Approximation of Most Probable Explanations in Bayesian Networks.- Argumentation Based Dynamic Multiple Criteria Decision Making.- Conditional Beliefs in a Bipolar Framework.- Detecting Marginal and Conditional Independencies between Events and Learning Their Causal Structure.- Measuring Incompleteness under Multi-valued Semantics by Partial MaxSAT Solvers.- On the Tree Structure Used by Lazy Propagation for Inference in Bayesian Networks.- Hierarchical Model for Rank Discrimination Measures.- Extreme Points of the Credal Sets Generated by Elementary Comparative Probabilities.- MCMC Estimation of Conditional Probabilities in Probabilistic Programming Languages.- Sorted-Pareto Dominance and Qualitative Notions of Optimality.- A First-Order Dynamic Probability Logic.- Selecting Source Behavior in Information Fusion on the Basis of Consistency and Specificity.- On the Problem of Reversing Relational Inductive Knowledge Representation.- Analogical Proportions and Multiple-Valued Logics.- Chain Graph Interpretations and Their Relations.- On the Plausibility of Abstract Arguments.
International Journal of Intelligent Systems, 1998
The concept of informational independence plays a key role in most knowledge-based systems. J. Pe... more The concept of informational independence plays a key role in most knowledge-based systems. J. Pearl and his co-researchers have analysed the basic properties of the concept and have formulated an axiomatic system for informational independence. This axiomatic system focuses on independences among mutually disjoint sets of variables. We show that in the context of probabilistic independence a focus on disjoint sets of variables can hide various interesting properties. To capture these properties, we enhance Pearl's axiomatic system with two additional axioms. We investigate the set of models of the thus enhanced system and show that it provides a better characterisation of the concept of probabilistic independence than Pearl's system does. In addition, we observe that both Pearl's axiomatic system and our enhanced system o er inference rules for deriving new independences from an initial set of independence statements and as such allow for a normal form for representing independence. We address the normal forms ensuing from the two axiomatic systems for informational independence.
Uncertainty in artificial intelligence : proceedings of the Twenty-third Conference (2007) : July 19-22, 2007, Vancouver, British Columbia
AUAI Press eBooks, 2007
Department of Computer Science [CS], 1989
The Centre for Mathematics and Computer Science is a research institute of the Stichting Mathemat... more The Centre for Mathematics and Computer Science is a research institute of the Stichting Mathematisch Centrum, which was founded on February 11, 1946, as a nonprofit institution aiming at the promotion of mathematics, computer science, and their applications. It is sponsored by the Dutch Govern- ment through the Netherlands Organization for the Advancement of Research (N.W.O.).
Evidence absorption : experiments on different classes of randomly generated belief networks
Unknown Publisher eBooks, 1994
ABSTRACT
Department of Computer Science [CS], 1987
Most expert knowledge is of an ill-defined and heuristic nature. Therefore, many present-day rule... more Most expert knowledge is of an ill-defined and heuristic nature. Therefore, many present-day rule-based expert systems include a mechanism for modelling and manipulating imprecise knowledge. For a long time, probability theory has been the primary quantitative approach for handling uncertainty. Other mathematical models of uncertainty have been proposed during the last decade, several of which depart from probability theory. The certainty factor model proposed by the authors of the MYCIN system is an example of an ad hoe model. The aim in developing the model was primarily to develop a method that was of practical use. The certainty factor model is computationally simple, a property that has led to its considerable success. In this paper, we use so-called inference networks to demonstrate the application of the model in a rule-based top-down reasoning expert system. This approach enables us to show some inadequacies of the notational convention used by the creators of the model. We propose a syntactically correct formalism and use this formalism to discuss several properties of the model.
Utrecht University: Information and Computing Sciences eBooks, 1996
Probabilistic inference with a belief network in general is computationally expensive. Since the ... more Probabilistic inference with a belief network in general is computationally expensive. Since the concept of structural relevance provides for identifying parts of a belief network that are irrelevant to a context of interest, it allows for alleviating to some extent the computational burden of inference: inference can be restricted to the network's relevant part. The structurally relevant part of a belief network, however, is not static. It may c hange dynamically as reasoning progresses. We address the dynamics of structural relevance and introduce the concept of an independence projection to capture these dynamics.
Congress of the Italian Association for Artificial Intelligence, Sep 14, 1999
arXiv (Cornell University), Jun 13, 2012
While known algorithms for sensitivity analysis and parameter tuning in probabilistic networks ha... more While known algorithms for sensitivity analysis and parameter tuning in probabilistic networks have a running time that is exponential in the size of the network, the exact computational complexity of these problems has not been established as yet. In this paper we study several variants of the tuning problem and show that these problems are NPPP-complete in general. We further show that the problems remain NP-complete or PP-complete, for a number of restricted variants. These complexity results provide insight in whether or not recent achievements in sensitivity analysis and tuning can be extended to more general, practicable methods.
Proceedings of the 12th European conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty
Invited Papers.- Graphical Models as Languages for Computer Assisted Diagnosis and Decision Makin... more Invited Papers.- Graphical Models as Languages for Computer Assisted Diagnosis and Decision Making.- Planning with Uncertainty and Incomplete Information.- What&amp;amp;amp;amp;#39;s Your Preference? And How to Express and Implement It in Logic Programming!.- Contributed Papers.- On Preference Representation on an Ordinal Scale.- Rule-Based Decision Support in Multicriteria Choice and Ranking.- Propositional Distances and Preference Representation.- Contributed Papers.- Value Iteration over Belief Subspace.- Space-Progressive Value Iteration: An Anytime Algorithm for a Class of POMDPs.- Contributed Papers.- Reasoning about Intentions in Uncertain Domains.- Troubleshooting with Simultaneous Models.- A Rational Conditional Utility Model in a Coherent Framework.- Contributed Papers.- Probabilistic Reasoning as a General Unifying Tool.- An Operational View of Coherent Conditional Previsions.- Contributed Papers.- Decomposition of Influence Diagrams.- Mixtures of Truncated Exponentials in Hybrid Bayesian Networks.- Importance Sampling in Bayesian Networks Using Antithetic Variables.- Using Recursive Decomposition to Construct Elimination Orders, Jointrees, and Dtrees.- Caveats For Causal Reasoning With Equilibrium Models.- Contributed Papers.- Supporting Changes in Structure in Causal Model Construction.- The Search of Causal Orderings: A Short Cut for Learning Belief Networks.- Stochastic Local Algorithms for Learning Belief Networks: Searching in the Space of the Orderings.- An Empirical Investigation of the K2 Metric.- Contributed Papers.- Sequential Valuation Networks: A New Graphical Technique for Asymmetric Decision Problems.- A Two-Steps Algorithm for Min-Based Possibilistic Causal Networks.- Contributed Papers.- Computing Intervals of Probabilities with Simulated Annealing and Probability Trees.- Probabilistic Logic under Coherence, Model-Theoretic Probabilistic Logic, and Default Reasoning.- Contributed Papers.- Belief Functions with Partially Ordered Values.- Dempster Specialization Matrices and the Combination of Belief Functions.- On the Conceptual Status of Belief Functions with Respect to Coherent Lower Probabilities.- About Conditional Belief Function Independence.- The Evaluation of Sensors&amp;amp;amp;amp;#39; Reliability and Their Tuning for Multisensor Data Fusion within the Transferable Belief Model.- Coarsening Approximations of Belief Functions.- Contributed Papers.- Label Semantics: A Formal Framework for Modeling with Words.- Reasoning about Knowledge Using Rough Sets.- Contributed Papers.- The Capacity of a Possibilistic Channel.- New Semantics for Quantitative Possibility Theory.- Bridging logical, comparative and graphical possibilistic representation frameworks.- Contributed Papers.- Ellipse fitting with uncertainty and fuzzy decision stage for detection. Application in videomicroscopy..- Probabilistic Modelling for Software Quality Control.- Spatial Information Revision: A Comparison between 3 Approaches.- Contributed Papers.- Social Choice, Merging, and Elections.- Data merging: Theory of Evidence vs knowledge-bases merging operators.- Contributed Papers.- A Priori Revision.- Some Operators for Iterated Revision.- On Computing Solutions to Belief Change Scenarios.- &amp;amp;amp;amp;quot;Not impossible&amp;amp;amp;amp;quot; vs. &amp;amp;amp;amp;quot;guaranteed possible&amp;amp;amp;amp;quot; in fusion and revision.- General Preferential Entaulments as Circumscriptions.- Contributed Papers.- A Semantic Tableau Version of First-Order Quasi-Classical Logic.- On Anytime Coherence-Based Reasoning.- Resolving Conflicts between Beliefs, Obligations, Intentions, and Desires.- Contributed Papers.- Comparing a Pair-wise Compatibility Heuristic and Relaxed Stratification: Some Preliminary Results.- How to Reason Credulously and Skeptically within a Single Extension.- Contributed Papers.- Handling Conditionals Adequately in Uncertain Reasoning.- Rankings We Prefer.- Contributed Papers.- Formalizing Human Uncertain Reasoning with Default Rules: A Psychological Conundrum and a Pragmatic Suggestion.- Statistical Information, Uncertainty, and Bayes&amp;amp;amp;amp;#39; Theorem: Some Applications in Experimental Psychology.- Polymorphism of Human Judgment under Uncertainty.- How to Doubt about a Conditional.- Contributed Papers.- Dialectical Proof Theories for the Credulous Preferred Semantics of Argumentation Frameworks.- Argumentation and Qualitative Probabilistic Reasoning Using the Kappa Calculus.- Contributed Papers.- Importance Measures from Reliability Theory for Probabilistic Assumption-Based Reasoning.- Ramification in the Normative Method of Causality.- Simultaneous Events: Conflicts and Preferences.- Orthogonal Relations for Reasoning about Abstract Events.- Contributed Papers.- Explanatory Relations Based on Mathematical Morphology.- Monotonic and Residuated Logic Programs.- Contributed Papers.- A Proof Procedure for Possibilistic Logic Programming with Fuzzy…
PROMUNDI: Probabilistic Multi-knowledge Networks for Diagnosis
International Symposium on Imprecise Probabilities and Their Applications, 2019
The Bayesian Dirichlet equivalent uniform (BDeu) function is a popular score to evaluate the good... more The Bayesian Dirichlet equivalent uniform (BDeu) function is a popular score to evaluate the goodness of a Bayesian network structure given complete categorical data. Despite its interesting properties, such as likelihood equivalence, it does require a prior expressed via a user-defined parameter known as Equivalent Sample Size (ESS), which significantly affects the final structure. We study conditions to obtain prior independence in BDeu-based structure learning. We show in experiments that the amount of data needed to render the learning robust to different ESS values is prohibitively large, even in big data times.