#### DMCA

## Markov Logic Networks (2006)

### Cached

### Download Links

- [www.cs.umass.edu]
- [www.cs.utexas.edu]
- [people.cs.umass.edu]
- [www.cs.washington.edu]
- [alchemy.cs.washington.edu]
- [research.microsoft.com]
- [ai.cs.washington.edu]
- [research.microsoft.com]
- [research.microsoft.com]
- [www.cs.washington.edu]
- [www.ics.uci.edu]
- [www.cs.washington.edu]
- [homes.cs.washington.edu]
- [homes.cs.washington.edu]
- [homes.cs.washington.edu]
- [eksl.isi.edu]
- [www.lirmm.fr]
- [homes.cs.washington.edu]
- [homes.cs.washington.edu]
- [homes.cs.washington.edu]
- [www.lirmm.fr]
- DBLP

### Other Repositories/Bibliography

Venue: | MACHINE LEARNING |

Citations: | 816 - 39 self |

### Citations

8904 |
Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference
- Pearl
- 1988
(Show Context)
Citation Context ...ections for future work (Section 10). 2. Markov Networks A Markov network (also known as Markov random field) is a model for the joint distribution of a set of variables X = (X1, X2, . . . , Xn) ∈ X (=-=Pearl, 1988-=-). It is composed of an undirected graph G and a set of potential functions φk. The graph has a node for each variable, and the model has a potential function for each clique in the graph. A potential... |

3613 |
Social Network Analysis: Methods and Applications
- Wasserman, Faust
- 1994
(Show Context)
Citation Context ...friends tend to have similar smoking habits (Lloyd-Richardson et al., 2002). In fact, an MLN like the one in Table I succinctly represents a type of model that is a staple of social network analysis (=-=Wasserman & Faust, 1994-=-). It is easy to see that MLNs subsume essentially all propositional probabilistic models, as detailed below. PROPOSITION 4.2. Every probability distribution over discrete or finiteprecision numeric v... |

3325 | Numerical optimization
- Nocedal, Wright
- 2000
(Show Context)
Citation Context ...nnot be computed in closed form, but, because the log-likelihood is a concave function of the weights, they can be found efficiently using standard gradientbased or quasi-Newton optimization methods (=-=Nocedal & Wright, 1999-=-). Another alternative is iterative scaling (Della Pietra et al., 1997). Features can also be learned from data, for example by greedily constructing conjunctions of atomic features (Della Pietra et a... |

2082 |
W.: Foundations of Logic Programming
- Lloyd
- 1984
(Show Context)
Citation Context ...re desirable properties. The most widely-used restriction is to Horn clauses, which are clauses containing at most one positive literal. The Prolog programming language is based on Horn clause logic (=-=Lloyd, 1987-=-). Prolog programs can be learned from databases by searching for Horn clauses that (approximately) hold in the data; this is studied in the field of inductive logic programming (ILP) (Lavrač & Dˇzero... |

1194 | Foundations of Inductive Logic Programming
- Muggleton
- 1995
(Show Context)
Citation Context ...binary clique between each predictor and the response. Noisy OR can similarly be represented with a linear number of parents. 9.3. OTHER LOGIC PROGRAMMING APPROACHES Stochastic logic programs (SLPs) (=-=Muggleton, 1996-=-; Cussens, 1999) are a combination of logic programming and log-linear models. Puech and Muggleton (2003) showed that SLPs are a special case of KBMC, and thus they can be converted into MLNs in the s... |

1158 | Learning Bayesian networks: the combination of knowledge and statistical data
- Heckerman, Geiger, et al.
- 1995
(Show Context)
Citation Context ... (for the All Info case) were discretized into five equal-frequency bins (based on the training set). We used two propositional learners: Naive Bayes (Domingos & Pazzani, 1997) and Bayesian networks (=-=Heckerman et al., 1995-=-) with structure mln.tex; 26/01/2006; 19:24; p.20sMarkov Logic Networks 21 and parameters learned using the VFBN2 algorithm (Hulten & Domingos, 2002) with a maximum of four parents per node. The order... |

1106 |
A machine-oriented logic based on the resolution principle.
- Robinson
- 1965
(Show Context)
Citation Context ...rst-order logic can be converted to clausal form using a mechanical sequence of steps. 1 Clausal form is used in resolution, a sound and refutation-complete inference procedure for first-order logic (=-=Robinson, 1965-=-). 1 This conversion includes the removal of existential quantifiers by Skolemization, which is not sound in general. However, in finite domains an existentially quantified formula can simply be repla... |

872 | Constraint logic programming, in:
- Jaffar, Lassez
- 1987
(Show Context)
Citation Context ...an be built up from classifiers (e.g., (Heckerman et al., 2000)), but this would be a significant extension of MACCENT. mln.tex; 26/01/2006; 19:24; p.34sMarkov Logic Networks 35 ues during inference (=-=Laffar & Lassez, 1987-=-). Probabilistic CLP generalizes SLPs to CLP (Riezler, 1998), and CLP(BN) combines CLP with Bayesian networks (Santos Costa et al., 2003). Unlike in MLNs, constraints in CLP(BN ) are hard (i.e., they ... |

856 |
Markov chain Monte Carlo in Practice
- Gilks, Richardson, et al.
- 1996
(Show Context)
Citation Context ...LNs will take advantage of this. Inference in Markov networks is #P-complete (Roth, 1996). The most widely used method for approximate inference in Markov networks is Markov chain Monte Carlo (MCMC) (=-=Gilks et al., 1996-=-), and in particular Gibbs sampling, which proceeds by sampling each variable in turn given its Markov blanket. (The Markov blanket of a node is the minimal set of nodes that renders it independent of... |

818 | On the optimality of the simple Bayesian classifier under zero-one loss.
- Domingos, Pazzani
- 1997
(Show Context)
Citation Context ... 28 order-1 attributes and 120 order-2 attributes (for the All Info case) were discretized into five equal-frequency bins (based on the training set). We used two propositional learners: Naive Bayes (=-=Domingos & Pazzani, 1997-=-) and Bayesian networks (Heckerman et al., 1995) with structure mln.tex; 26/01/2006; 19:24; p.20sMarkov Logic Networks 21 and parameters learned using the VFBN2 algorithm (Hulten & Domingos, 2002) wit... |

789 | Probability and Statistics - DeGroot - 1986 |

780 | On the limited memory BFGS method for large scale optimization
- Liu, Nocedal
- 1989
(Show Context)
Citation Context ...ed, and similarly for ni(x [Xl=1]). Computing this expression (or Equation 7) does not require inference over the model. We optimize the pseudo-log-likelihood using the limited-memory BFGS algorithm (=-=Liu & Nocedal, 1989-=-). The computation can be made more efficient in several ways: − The sum in Equation 8 can be greatly sped up by ignoring predicates that do not appear in the ith formula. − The counts ni(x), ni(x [Xl... |

670 | Inducing features of random fields.
- Pietra, Pietra, et al.
- 1995
(Show Context)
Citation Context ...cave function of the weights, they can be found efficiently using standard gradientbased or quasi-Newton optimization methods (Nocedal & Wright, 1999). Another alternative is iterative scaling (Della =-=Pietra et al., 1997-=-). Features can also be learned from data, for example by greedily constructing conjunctions of atomic features (Della Pietra et al., 1997). 3. First-Order Logic A first-order knowledge base (KB) is a... |

660 | Discriminative training methods for hidden Markov models: Theory and experiments with perceptron algorithms.
- Collins
- 2002
(Show Context)
Citation Context ...e training of MLNs is straightforward (in fact, easier than the generative training used in this paper), and we have carried out successful preliminary experiments using a voted perceptron algorithm (=-=Collins, 2002-=-). RMNs use MAP estimation with belief propagation for inference, which makes learnmln.tex; 26/01/2006; 19:24; p.35s36 Richardson and Domingos ing quite slow, despite the simplified discriminative set... |

613 | Learning Probabilistic Relational Models
- Friedman, Getoor, et al.
- 1999
(Show Context)
Citation Context ...raints in CLP(BN ) are hard (i.e., they cannot be violated; rather, they define the form of the probability distribution). 9.4. PROBABILISTIC RELATIONAL MODELS Probabilistic relational models (PRMs) (=-=Friedman et al., 1999-=-) are a combination of frame-based systems and Bayesian networks. PRMs can be converted into MLNs by defining a predicate S(x, v) for each (propositional or relational) attribute of each class, where ... |

572 | A limited memory algorithm for bound constrained optimization. - Byrd, Lu, et al. - 1995 |

494 | Probabilistic logic. - Nilsson - 1986 |

474 | Generalized belief propagation. In
- Yedidia, Freeman, et al.
- 2001
(Show Context)
Citation Context ...l probabilities are computed by running the Gibbs sampler with the conditioning variables clamped to their given values. Another popular method for inference in Markov networks is belief propagation (=-=Yedidia et al., 2001-=-). Maximum-likelihood or MAP estimates of Markov network weights cannot be computed in closed form, but, because the log-likelihood is a concave function of the weights, they can be found efficiently ... |

453 | Enhanced hypertext categorization using hyperlinks - Chakrabarti, Dom, et al. - 1998 |

415 | Discriminative probabilistic models for relational data
- Taskar, Abbeel, et al.
- 2002
(Show Context)
Citation Context ...In particular, potential functions must be conditional probabilities, and the directed graph must have no cycles. The latter condition is particularly troublesome to enforce in relational extensions (=-=Taskar et al., 2002-=-). (4) mln.tex; 29/07/2005; 14:21; p.13s14 Richardson and Domingos appearing in KB or F . The question is answered by computing P (F |LKB, CKB,F ) by Equation 4, with F2 = True. Computing Equation 4 d... |

385 |
Statistical analysis of non-lattice data
- Besag
- 1975
(Show Context)
Citation Context ...chains yielded poor results. A more efficient alternative, widely used in areas like spatial statistics, social network modeling and language processing, is to optimize instead the pseudo-likelihood (=-=Besag, 1975-=-) P ∗ w(X =x) = n� Pw(Xl =xl|MBx(Xl)) (7) l=1 where MBx(Xl) is the state of the Markov blanket of Xl in the data. The gradient of the pseudo-log-likelihood is mln.tex; 26/01/2006; 19:24; p.16s∂ ∂wi lo... |

328 | Probabilistic horn abduction and Bayesian networks,
- Poole
- 1993
(Show Context)
Citation Context ... over predicates; the latter have to be obtained by marginalization. Similar remarks apply to a number of other representations that are essentially equivalent to SLPs, like independent choice logic (=-=Poole, 1993-=-) and PRISM (Sato & Kameya, 1997). MACCENT (Dehaspe, 1997) is a system that learns log-linear models with first-order features; each feature is a conjunction of a class and a Prolog query (clause with... |

314 | An analysis of first-order logics of probability, - Halpern - 1990 |

303 | The state of record linkage and current research problems
- Winkler
- 1999
(Show Context)
Citation Context ...uplication, and others) is the problem of determining which records in a database refer to the same real-world entity (e.g., which entries in a bibliographic database represent the same publication) (=-=Winkler, 1999-=-). This problem is of crucial importance to many companies, government agencies, and large-scale scientific projects. One way to represent it in MLNs is by removing the unique names assumption as desc... |

293 | Efficient identification of Web communities. In: - Flake, Giles |

289 | On the hardness of approximate reasoning.
- Roth
- 1996
(Show Context)
Citation Context ...llowing for a more compact representation than the potential-function form, particularly when large cliques are present. MLNs will take advantage of this. Inference in Markov networks is #P-complete (=-=Roth, 1996-=-). The most widely used method for approximate inference in Markov networks is Markov chain Monte Carlo (MCMC) (Gilks et al., 1996), and in particular Gibbs sampling, which proceeds by sampling each v... |

285 |
Constrained Monte Carlo Maximum Likelihood for Dependent Data.”
- Geyer, Thompson
- 1992
(Show Context)
Citation Context ...ion methods also require computing the log-likelihood itself (Equation 3), and thus the partition function Z. This can be done approximately using a Monte Carlo maximum likelihood estimator (MC-MLE) (=-=Geyer & Thompson, 1992-=-). However, in our experiments the Gibbs sampling used to compute the MC-MLEs and gradients did not converge in reasonable time, and using the samples from the unconverged chains yielded poor results.... |

284 | Representing and Reasoning with Probabilistic Knowledge: A Logical Approach to Probabilities. - Bacchus - 1990 |

276 | Operations for learning with graphical models",
- Buntine
- 1994
(Show Context)
Citation Context ...ution of a Gibbs sampler operating on it (Heckerman et al., 2000). 9.8. PLATES AND PROBABILISTIC ER MODELS Large graphical models with repeated structure are often compactly represented using plates (=-=Buntine, 1994-=-). MLNs subsume plates as a representation language. In addition, they allow individuals and their relations to be explicitly represented (see Cussens (2003)), and context-specific independencies to b... |

257 |
Logical foundations of artificial intelligence
- Genesereth, Nilsson
- 1987
(Show Context)
Citation Context ...ple by greedily constructing conjunctions of atomic features (Della Pietra et al., 1997). 3. First-Order Logic A first-order knowledge base (KB) is a set of sentences or formulas in firstorder logic (=-=Genesereth & Nilsson, 1987-=-). Formulas are constructed using four types of symbols: constants, variables, functions, and predicates. Constant symbols represent objects in the domain of interest (e.g., people: Anna, Bob, Chris, ... |

208 | Dependency networks for inference, collaborative filtering, and data visualization.
- Heckerman, Chickering, et al.
- 2000
(Show Context)
Citation Context ...programming (CLP) is an extension of logic programming where variables are constrained instead of being bound to specific val4 Conversely, joint distributions can be built up from classifiers (e.g., (=-=Heckerman et al., 2000-=-)), but this would be a significant extension of MACCENT. mln.tex; 26/01/2006; 19:24; p.34sMarkov Logic Networks 35 ues during inference (Laffar & Lassez, 1987). Probabilistic CLP generalizes SLPs to ... |

199 | Clausal discovery. - Raedt, Dehaspe - 1997 |

191 | First-order probabilistic inference. - Poole - 2003 |

185 | Knowledge-based artificial neural networks.
- Towell, Shavlik
- 1994
(Show Context)
Citation Context ...pplicable to MLNs. MLNs have some interesting similarities with the KBANN system, which converts a propositional Horn KB into a neural network and uses backpropagation to learn the network’s weights (=-=Towell & Shavlik, 1994-=-). More generally, MLNs can be viewed as an extension to probability estimation of a long line of work on knowledge-intensive learning (e.g., Bergadano and Giordana (1988); Pazzani and Kibler (1992); ... |

181 |
Generalization of the Fortuin-Kasteleyn-Swendsen-Wang representation and Monte Carlo algorithm,
- Edwards, Sokal
- 1988
(Show Context)
Citation Context ...gence threshold and limited the training process to less than ten hours. With a better choice of initial state, approximate counting, and improved MCMC techniques such as the Swendsen-Wang algorithm (=-=Edwards & Sokal, 1988-=-), MC-MLE may become practical, but it is not a viable option for training in the current version. (Notice that during learning MCMC is performed over the full ground network, which is too large to ap... |

154 | The utility of knowledge in inductive learning, - Pazzani, Kibler - 1992 |

151 | Inductive Logic Programming: Techniques and Applications. Ellis Horwood. - Lavrac, Dzeroski - 1994 |

129 | Theory refinement combining analytical and empirical methods. - Ourston, Mooney - 1994 |

122 | From Statistical Knowledge Bases to Degrees of Belief. - Bacchus, Grove, et al. - 1996 |

98 | Answering queries from context-sensitive probabilistic knowledge bases,
- Ngo, Haddawy
- 1997
(Show Context)
Citation Context ...ncer-free smokers in it increases. 9.2. KNOWLEDGE-BASED MODEL CONSTRUCTION Knowledge-based model construction (KBMC) is a combination of logic programming and Bayesian networks (Wellman et al., 1992; =-=Ngo & Haddawy, 1997-=-; Kersting & De Raedt, 2001). As in MLNs, nodes in KBMC represent ground atoms. Given a Horn KB, KBMC answers a query by finding all possible backward-chaining proofs of the query and evidence atoms f... |

83 | Towards combining inductive logic programming with Bayesian networks. - Kersting, Raedt - 2001 |

81 | L-BFGS-B: Algorithm 778: L-BFGSB, FORTRAN routines for large scale bound constrained optimization, - Zhu, Byrd, et al. - 1997 |

79 |
From Knowledge Bases to Decision Models,
- Wellman, Breese, et al.
- 1992
(Show Context)
Citation Context ...unding them) is an important direction for future work (see Jaeger (2000) and Poole (2003) for initial results). The algorithm proceeds in two phases, analogous to knowledge-based model construction (=-=Wellman et al., 1992-=-). The first phase returns the minimal subset M of the ground Markov network required to compute P (F1|F2, L, C). The algorithm for this is shown in Table III. The size of the network returned may be ... |

64 | A general stochastic approach to solving problems with hard and soft constraints.
- Kautz, Selman, et al.
- 1997
(Show Context)
Citation Context ...ach run from a mode found using MaxWalkSat, a local search algorithm for the weighted satisfiability problem (i.e., finding a truth assignment that maximizes the sum of weights of satisfied clauses) (=-=Kautz et al., 1997-=-). When there are hard constraints (clauses with infinite weight), MaxWalkSat finds regions that satisfy them, and the Gibbs sampler then samples from these regions to obtain probability estimates. 6.... |

64 | Multi-Relational Record Linkage.
- Domingos, Domingos
- 2004
(Show Context)
Citation Context ... and y are records and fi(x) is a function returning the value of the ith field of record x. We have successfully applied this approach to de-duplicating the Cora database of computer science papers (=-=Parag & Domingos, 2004-=-). Because it allows information to propagate from one match decision (i.e., one grounding of x = y) to another via fields that appear in both pairs of records, it effectively performs collective obje... |

63 | Collective classification with relational dependency networks.
- Neville, Jensen
- 2003
(Show Context)
Citation Context ... a discriminatively-trained MLN. 5 9.7. RELATIONAL DEPENDENCY NETWORKS In a relational dependency network (RDN), each node’s probability conditioned on its Markov blanket is given by a decision tree (=-=Neville & Jensen, 2003-=-). Every RDN has a corresponding MLN in the same way that every dependency network has a corresponding Markov network, given by the stationary distribution of a Gibbs sampler operating on it (Heckerma... |

63 | CLP(BN ): Constraint logic programming for probabilistic knowledge.
- Costa, Page, et al.
- 2003
(Show Context)
Citation Context ...; 19:24; p.34sMarkov Logic Networks 35 ues during inference (Laffar & Lassez, 1987). Probabilistic CLP generalizes SLPs to CLP (Riezler, 1998), and CLP(BN) combines CLP with Bayesian networks (Santos =-=Costa et al., 2003-=-). Unlike in MLNs, constraints in CLP(BN ) are hard (i.e., they cannot be violated; rather, they define the form of the probability distribution). 9.4. PROBABILISTIC RELATIONAL MODELS Probabilistic re... |

51 | Dynamic probabilistic relational models. - Sanghai, Domingos, et al. - 2003 |

49 |
PRISM: A symbolic-statistical modeling language.
- Sato, Kameya
- 1997
(Show Context)
Citation Context ...tter have to be obtained by marginalization. Similar remarks apply to a number of other representations that are essentially equivalent to SLPs, like independent choice logic (Poole, 1993) and PRISM (=-=Sato & Kameya, 1997-=-). MACCENT (Dehaspe, 1997) is a system that learns log-linear models with first-order features; each feature is a conjunction of a class and a Prolog query (clause with empty head). A key difference b... |

48 | Approximate inference for first-order probabilistic languages - Pasula, Russell - 2001 |

45 | Building large knowledge bases by mass collaboration, in:
- Richardson, Domingos
- 2003
(Show Context)
Citation Context ... An MLN can also be obtained by merging several KBs, even if they are partly incompatible. This is potentially useful in areas like the Semantic Web (Berners-Lee et al., 2001) and mass collaboration (=-=Richardson & Domingos, 2003-=-). It is interesting to see a simple example of how MLNs generalize firstorder logic. Consider an MLN containing the single formula ∀x R(x) ⇒ S(x) with weight w, and C = {A}. This leads to four possib... |

41 | Probabilistic Constraint Logic Programming.
- Riezler
- 1998
(Show Context)
Citation Context ...ut this would be a significant extension of MACCENT. mln.tex; 26/01/2006; 19:24; p.34sMarkov Logic Networks 35 ues during inference (Laffar & Lassez, 1987). Probabilistic CLP generalizes SLPs to CLP (=-=Riezler, 1998-=-), and CLP(BN) combines CLP with Bayesian networks (Santos Costa et al., 2003). Unlike in MLNs, constraints in CLP(BN ) are hard (i.e., they cannot be violated; rather, they define the form of the pro... |

36 | Maximum entropy modeling with clausal constraints
- Dehaspe
- 1997
(Show Context)
Citation Context ...e logic programming (ILP) techniques can be used to learn additional clauses, refine the ones already in the MLN, or learn an MLN from scratch. We use the CLAUDIEN system for this purpose (De Raedt & =-=Dehaspe, 1997-=-). Unlike most other ILP systems, which learn only Horn clauses, CLAUDIEN is able to learn arbitrary first-order clauses, making it well suited to MLNs. Also, by constructing a particular language bia... |

36 | Blog: Relational modeling with unknown objects. In: - Milch, Marthi, et al. - 2004 |

35 | Mining Complex Models from Arbitrarily Large Databases in Constant Time.
- Hulten, Domingos
- 2002
(Show Context)
Citation Context ...es (Domingos & Pazzani, 1997) and Bayesian networks (Heckerman et al., 1995) with structure mln.tex; 26/01/2006; 19:24; p.20sMarkov Logic Networks 21 and parameters learned using the VFBN2 algorithm (=-=Hulten & Domingos, 2002-=-) with a maximum of four parents per node. The order-2 attributes helped the naive Bayes classifier but hurt the performance of the Bayesian network classifier, so below we report results using the or... |

33 | Loglinear models for first-order probabilistic reasoning
- CUSSENS
- 1999
(Show Context)
Citation Context ...ween each predictor and the response. Noisy OR can similarly be represented with a linear number of parents. 9.3. OTHER LOGIC PROGRAMMING APPROACHES Stochastic logic programs (SLPs) (Muggleton, 1996; =-=Cussens, 1999-=-) are a combination of logic programming and log-linear models. Puech and Muggleton (2003) showed that SLPs are a special case of KBMC, and thus they can be converted into MLNs in the same way. Like M... |

33 | Structural logistic regression for link analysis
- Popescul, Popescul, et al.
- 2003
(Show Context)
Citation Context ...ified discriminative setting; maximizing the pseudo-likelihood of the query variables may be a more effective alternative. 9.6. STRUCTURAL LOGISTIC REGRESSION In structural logistic regression (SLR) (=-=Popescul & Ungar, 2003-=-), the predictors are the output of SQL queries over the input data. Just as a logistic regression model is a discriminatively-trained Markov network, an SLR model is a discriminatively-trained MLN. 5... |

31 | Probabilistic entityrelationship models, PRMs, and plate models,” - Heckerman, Meek, et al. - 2007 |

19 | A comparison of stochastic logic programs and Bayesian logic programs. - Puech, Muggleton - 2003 |

17 | On the complexity of inference about probabilistic relational models, - Jaeger - 2000 |

16 |
Differentiating stages of smoking intensity among adolescents: stage-specific psychological and social influences.
- Lloyd-Richardson, Papandonatos, et al.
- 2002
(Show Context)
Citation Context ... but capture useful information on friendships and smoking habits, when viewed as features of a Markov network. For example, it is well known that teenage friends tend to have similar smoking habits (=-=Lloyd-Richardson et al., 2002-=-). In fact, an MLN like the one in Table I succinctly represents a type of model that is a staple of social network analysis (Wasserman & Faust, 1994). It is easy to see that MLNs subsume essentially ... |

15 | Reasoning about infinite random structures with relational Bayesian networks, in: - Jaeger - 1998 |

14 | Feature extraction languages for propositionalized relational learning - Roth |

10 | Maximum entropy probabilistic logic - PASKIN |

5 | Aknowledgeintensive approach to concept induction - Bergadano, Giordana - 1988 |

3 |
to Other Fields
- Dietterich, Getoor, et al.
- 2003
(Show Context)
Citation Context ...Many (if not most) applications require both. Interest in this problem has grown in recent years due to its relevance to statistical relational learning (Getoor & Jensen, 2000; Getoor & Jensen, 2003; =-=Dietterich et al., 2003-=-), also known as multi-relational data mining (Dˇzeroski & De Raedt, 2003; Dˇzeroski et al., 2002; Dˇzeroski et al., 2003; Dˇzeroski & Blockeel, 2004). Current proposals typically focus on combining p... |

2 | Individuals, relations and structures in probabilistic models - Cussens - 2003 |

2 | Special issue on multi-relational data mining: The current frontiers - Dˇzeroski, Raedt - 2003 |

1 |
mln.tex; 26/01/2006; 19:24; p.42 Markov Logic Networks 43
- Taskar, Abbeel, et al.
- 2002
(Show Context)
Citation Context ...In particular, potential functions must be conditional probabilities, and the directed graph must have no cycles. The latter condition is particularly troublesome to enforce in relational extensions (=-=Taskar et al., 2002-=-). mln.tex; 26/01/2006; 19:24; p.12sMarkov Logic Networks 13 The size of ground Markov networks can be vastly reduced by having typed constants and variables, and only grounding variables to constants... |

1 | mln.tex; 29/07/2005; 14:21; p.38 Markov Logic Networks 39 - Bacchus, Grove, et al. - 1996 |

1 | on Statistical Relational Learning and its Connections to Other Fields
- Dietterich, Getoor, et al.
- 2003
(Show Context)
Citation Context ...Many (if not most) applications require both. Interest in this problem has grown in recent years due to its relevance to statistical relational learning (Getoor & Jensen, 2000; Getoor & Jensen, 2003; =-=Dietterich et al., 2003-=-), also known as multi-relational data mining (Dˇzeroski & De Raedt, 2003; Dˇzeroski et al., 2002; Dˇzeroski et al., 2003; Dˇzeroski & Blockeel, 2004). Current proposals typically focus on combining p... |