Structured machine learning: the next ten years

Dietterich, Thomas G.; Domingos, Pedro; Getoor, Lise; Muggleton, Stephen; Tadepalli, Prasad

doi:10.1007/s10994-008-5079-1

Cited by 95 publications

(67 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Structured machine learning (Dietterich et al 2008) has received an increased amount of interest in the recent years for several reasons like allowing to handle complex data in a natural way (as illustrated by the success of these techniques in biomedical fields), or sophisticated forms of inference. Most recent work on similarity assessment for structured representations follows the idea of "hierarchical aggregation", in which to compute the similarity between two instances, each instance is seen as an object with a set of features and the similarity is computed as an aggregation of the similarity of the values in the features.…”

mentioning

confidence: 99%

Similarity measures over refinement graphs

Ontañón

Plaza

2011

Mach Learn

View full text Add to dashboard Cite

Similarity also plays a crucial role in support vector machines. Similarity assessment plays a key role in lazy learning methods such as k-nearest neighbor or case-based reasoning. In this paper we will show how refinement graphs, that were originally introduced for inductive learning, can be employed to assess and reason about similarity. We will define and analyze two similarity measures, S λ and S π , based on refinement graphs. The anti-unification-based similarity, S λ , assesses similarity by finding the anti-unification of two instances, which is a description capturing all the information common to these two instances. The property-based similarity, S π , is based on a process of disintegrating the instances into a set of properties, and then analyzing these property sets. Moreover these similarity measures are applicable to any representation language for which a refinement graph that satisfies the requirements we identify can be defined. Specifically, we present a refinement graph for feature terms, in which several languages of increasing expressiveness can be defined. The similarity measures are empirically evaluated on relational data sets belonging to languages of different expressiveness.

show abstract

mentioning

confidence: 99%

Similarity measures over refinement graphs

Ontañón

Plaza

2011

Mach Learn

View full text Add to dashboard Cite

show abstract

“…In [4], for example, we can find an extensive description of the current and emerging trends in the so-called 'structured machine learning' where the authors propose to go beyond supervised learning and inference, and consider decision-theoretic planning and reinforcement learning in relational and first-order settings.…”

Section: Previous Workmentioning

confidence: 99%

Learning with Configurable Operators and RL-Based Heuristics

Martínez-Plumed

Ferri

Hernández-Orallo

et al. 2013

New Frontiers in Mining Complex Patterns

View full text Add to dashboard Cite

Abstract. In this paper, we push forward the idea of machine learning systems for which the operators can be modified and finetuned for each problem. This allows us to propose a learning paradigm where users can write (or adapt) their operators, according to the problem, data representation and the way the information should be navigated. To achieve this goal, data instances, background knowledge, rules, programs and operators are all written in the same functional language, Erlang. Since changing operators affect how the search space needs to be explored, heuristics are learnt as a result of a decision process based on reinforcement learning where each action is defined as a choice of operator and rule. As a result, the architecture can be seen as a 'system for writing machine learning systems' or to explore new operators.

show abstract

“…It seems that there is a general agreement among researchers, also expressed by Dietterich et al (2008), that (a) predicate invention is a key problem in ML, and (b) it is exceptionally hard due to its high combinatorial complexity, maybe too hard. The number of alternatives in which new predicates can be introduced is extremely explosive, and recognizing their (potential) benefits early in the search among all these alternatives is very hard.…”

Section: Whatever Happened To Predicate Invention?mentioning

confidence: 99%

“…A method to discover unknown relations from incomplete networks has been recently introduced in Inoue et al (2010) based on meta-level abduction, which infer missing rules, missing facts, and unknown causes that involve predicate invention in the form of existentially quantified hypotheses. This invention is also a realization of hidden object invention (Dietterich et al 2008), but lifting meta-level abduction to higher-order is also a promising method in this direction.…”

Section: The Engineering Of Ilp Systems: How?mentioning

confidence: 99%

ILP turns 20

et al. 2011

Self Cite

View full text Add to dashboard Cite

Inductive Logic Programming (ILP) is an area of Machine Learning which has now reached its twentieth year. Using the analogy of a human biography this paper recalls the development of the subject from its infancy through childhood and teenage years. We show how in each phase ILP has been characterised by an attempt to extend theory and implementations in tandem with the development of novel and challenging real-world applications. Lastly, by projection we suggest directions for research which will help the subject coming of age.

show abstract

Structured machine learning: the next ten years

Cited by 95 publications

References 57 publications

Similarity measures over refinement graphs

Similarity measures over refinement graphs

Learning with Configurable Operators and RL-Based Heuristics

ILP turns 20

Contact Info

Product

Resources

About