Filters








957 Hits in 8.4 sec

Defaults and Infinitesimals: Defeasible Inference by Nonarchimedean Entropy-Maximization [article]

Emil Weydert
2013 arXiv   pre-print
We develop a new semantics for defeasible inference based on extended probability measures allowed to take infinitesimal values, on the interpretation of defaults as generalized conditional probability  ...  constraints and on a preferred-model implementation of entropy maximization.  ...  [Leh 92, BCDLP 93], conditional entailment ll=cE [GP 92], maximum-entropy entailment II=ME [GMP 90, Gol 92] and random-worlds entailment II=Rw [BGHK 93]. maximum-entropy-based formalism for defeasible  ... 
arXiv:1302.4988v1 fatcat:q55wyeuvcberde6hkrp6zelqey

Representation Dependence in Probabilistic Inference

J. Y. Halpern, D. Koller
2004 The Journal of Artificial Intelligence Research  
Moreover, we show that representation independence is incompatible with even a weak default assumption of independence.  ...  In this paper, we formalize this notion and show that it is not a problem specific to maximum entropy.  ...  To make our discussion more concrete, we discuss this issue in one particular context: probabilistic inference.  ... 
doi:10.1613/jair.1292 fatcat:ucalmszeerdrrcw2tmf4m2udv4

PrASP Report [article]

Matthias Nickles
2016 arXiv   pre-print
PrASP is a research software which integrates non-monotonic reasoning based on Answer Set Programming (ASP), probabilistic inference and parameter learning.  ...  In particular, PrASP allows for ASP as well as First-Order Logic syntax, and for the annotation of formulas with point probabilities as well as interval probabilities.  ...  This obviously does not guarantee maximum entropy but merely aims at avoiding minimum entropy.  ... 
arXiv:1612.09591v1 fatcat:6znpsfmqs5gw5jnhjgdcxi2qny

Recent Advances in Querying Probabilistic Knowledge Bases

Stefan Borgwardt, İsmail İlkan Ceylan, Thomas Lukasiewicz
2018 Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence  
We give a survey on recent advances at the forefront of research on probabilistic knowledge bases for representing and querying large-scale automatically extracted data.  ...  We concentrate especially on increasing the semantic expressivity of formalisms for representing and querying probabilistic knowledge (i) by giving up the closed-world assumption, (ii) by allowing for  ...  Observe that this is fundamentally different from the probabilistic semantics above that are based on Markov logic networks or maximum-entropy models.  ... 
doi:10.24963/ijcai.2018/765 dblp:conf/ijcai/BorgwardtCL18 fatcat:o43b3gn6sbgydncx55ryiqftde

A Logic for Default Reasoning About Probabilities [chapter]

Manfred Jaeger
1994 Uncertainty Proceedings 1994  
Cross entropy minimization i s a k ey element in these semantics, the use of which is justi ed by showing that the resulting logic exhibits some very reasonable properties.  ...  Acknowledgement The author is greatful for some helpful remarks and suggestions received from an anonymous referee.  ...  Particularly, they contained a valuable clari cation regarding the interrelation of direct inference and Jeffrey's rule.  ... 
doi:10.1016/b978-1-55860-332-5.50049-3 fatcat:qosgbgrngfhs5o54optcjzhmte

Evaluation and Comparison Criteria for Approaches to Probabilistic Relational Knowledge Representation [chapter]

Christoph Beierle, Marc Finthammer, Gabriele Kern-Isberner, Matthias Thimm
2011 Lecture Notes in Computer Science  
, and three approaches based on the principle of maximum entropy.  ...  with probabilistic reasoning.  ...  A technical comparison with respect to default reasoning properties of the three approaches employing the principle of maximum entropy can be found in [12] .  ... 
doi:10.1007/978-3-642-24455-1_6 fatcat:orp6j3hmj5btnovnt4czh436xa

A Hybrid Approach to Inference in Probabilistic Non-Monotonic Logic Programming

Matthias Nickles, Alessandra Mileo
2015 International Conference on Logic Programming  
We present a probabilistic inductive logic programming framework which integrates non-monotonic reasoning, probabilistic inference and parameter learning.  ...  for adaptability with regard to different reasoning and learning tasks.  ...  formula f or ¬f into a disjunction f |¬ f , where ¬ stands for default negation.  ... 
dblp:conf/iclp/NicklesM15 fatcat:s5yiq73325endpnru5lctb5rxe

Representation Dependence in Probabilistic Inference [article]

Joseph Y. Halpern, Daphne Koller
2003 arXiv   pre-print
Moreover, we show that representation independence is incompatible with even a weak default assumption of independence.  ...  In this paper, we formalize this notion and show that it is not a problem specific to maximum entropy.  ...  For example, maximum entropy (or any inference procedure based on symmetry) will conclude Pr(p) = 1/2 from the empty knowledge base.  ... 
arXiv:cs/0312048v1 fatcat:k4lbvei7qrdizlwvcnqf6n3qme

A Logic for Default Reasoning About Probabilities [article]

Manfred Jaeger
2013 arXiv   pre-print
Cross entropy minimization is a key element in these semantics, the use of which is justified by showing that the resulting logic exhibits some very reasonable properties.  ...  Acknowledgement The author is greatful for some helpful remarks and suggestions received from an anonymous referee.  ...  Par ticularly, they contained a valuable clarification re garding the interrelation of direct inference and Jef frey's rule.  ... 
arXiv:1302.6822v1 fatcat:rnfpyqq6sng73n7rcbsjjhfldq

Conditional Logics and Conditional Reasoning: New Joint Perspectives (Dagstuhl Seminar 19032)

Guillaume Aucher, Paul Egré, Gabriele Kern-Isberner, Francesca Poggliesi, Michael Wagner
2019 Dagstuhl Reports  
In the last decades, with the emergence of artificial intelligence, a large number of logics called conditional logics have been introduced to model our conditional reasoning captured by socalled conditionals  ...  More recently, conditional reasoning has also come under scrutiny by psychologists, yet with more pragmatic and empirical considerations.  ...  Prepare a paper on the division of labor between semantics and pragmatics and the relation between inferential and difference-making conditionals. -Conditional  ... 
doi:10.4230/dagrep.9.1.47 dblp:journals/dagstuhl-reports/AucherEKP19 fatcat:vgv66achk5glbfzakqtf6kzcrq

From Statistical Knowledge Bases to Degrees of Belief [article]

Fahiem Bacchus, Adam Grove, Joseph Y. Halpern, Daphne Koller
2003 arXiv   pre-print
Our results show that a number of desiderata that arise in direct inference (reasoning from statistical information to conclusions about individuals) and default reasoning follow directly from the semantics  ...  It is able to integrate qualitative default reasoning with quantitative probabilistic reasoning by providing a language in which both types of information can be easily expressed.  ...  Shastri's result is based on maximum entropy.  ... 
arXiv:cs/0307056v1 fatcat:wxih2voyorcxvfuozljaapueta

Representing and Reasoning With Probabilistic Knowledge: A Bayesian Approach [article]

Marie desJardins
2013 arXiv   pre-print
knowledge to guide and constrain the learning process and for representing, reasoning with, and learning probabilistic knowledge.  ...  These theories are represented as conditional probability distributions.  ...  reason ing, including certain forms of default reasoning, it does not provide a representation for beliefs about rel evance, nor does it allow default assumptions such as independence or maximum entropy  ... 
arXiv:1303.1481v1 fatcat:jx6tmlbtwjbwxknxunom5eapha

Probabilistic Inductive Logic Programming Based on Answer Set Programming [article]

Matthias Nickles, Alessandra Mileo
2014 arXiv   pre-print
We propose a new formal language for the expressive representation of probabilistic knowledge based on Answer Set Programming (ASP).  ...  It allows for the annotation of first-order formulas as well as ASP rules and facts with probabilities and for learning of such weights from data (parameter estimation).  ...  maximum entropy (Thimm and Kern-Isberner 2012), in case multiple solutions exist 2 .  ... 
arXiv:1405.0720v1 fatcat:ahr4hjvqmzf6hdy5cj34gsfesa

On probabilistic inference in relational conditional logics

M. Thimm, G. Kern-Isberner
2012 Logic Journal of the IGPL  
Due to this principle, reasoning is performed based on the unique model of a knowledge base that has maximum entropy.  ...  The principle of maximum entropy has proven to be a powerful approach for commonsense reasoning in probabilistic conditional logics on propositional languages.  ...  Relational Conditional Logic and Maximum Entropy The approach of lifting inference in conditional logic based on the principle of maximum entropy to the first-order case has been previously investigated  ... 
doi:10.1093/jigpal/jzs010 fatcat:6i5ziij6ofaohm7z4lw4somnqa

From statistical knowledge bases to degrees of belief

Fahiem Bacchus, Adam J. Grove, Joseph Y. Halpern, Daphne Koller
1996 Artificial Intelligence  
It is able to integrate qualitative default reasoning with quantitative probabilistic reasoning by providing a language in which both types of information can be easily expressed.  ...  Our results show that a number of desiderata that arise in direct inference (reasoning * A preliminary version of this from statistical information to conclusions about individuals) and default reasoning  ...  Shastri's result is based on maximum entropy.  ... 
doi:10.1016/s0004-3702(96)00003-3 fatcat:s5ho47qjcrcl5afyiql5n4elb4
« Previous Showing results 1 — 15 out of 957 results