Propositionalization-based relational subgroup discovery with RSD

Železný, Filip; Lavrač, Nada

doi:10.1007/s10994-006-5834-0

Cited by 87 publications

(65 citation statements)

References 29 publications

(54 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The clusters of the best clusterings according to clustering quality and expert evaluation are further interpreted by the corresponding non-image data. For the interpretation of image clusters, we employ relational subgroup discovery (RSD) [10,17], an algorithm for finding interesting subgroups in data. In subgroup discovery, the goal is to find subgroup descriptions (typically conjunctions of attribute values as in rule learning) for which the distribution of examples with respect to a specified target variable is "unusual" compared to the overall target distribution.…”

Section: Methodsmentioning

confidence: 99%

“…In the second step, we explained the clusters by relating them to clinical and other non-image variables. To do so, we employed RSD [10,17], an algorithm for relational subgroup discovery, with the cluster membership of patients as the target variable. After extracting relevant information from 200 GB of data (removing duplicates, intermediate results, and incompletely processed images), we obtained a dataset comprising 10 GB of image data from 454 PETs, and 42 variables from clinical and demographical data organized in 11 relations of a relational database.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Interpreting PET Scans by Structured Patient Data: A Data Mining Case Study in Dementia Research

Hapfelmeier¹,

Schmidt²,

Mueller³

et al. 2008

2008 Eighth IEEE International Conference on Data Mining

View full text Add to dashboard Cite

One of the goals of medical research in the area of dementia is to correlate images of the brain with clinical tests. Our approach is to start with the images and explain the differences and commonalities in terms of the other variables. First, we cluster Positron emission tomography (PET) scans of patients to form groups sharing similar features in brain metabolism. To the best of our knowledge, it is the first time ever that clustering is applied to whole PET scans. Second, we explain the clusters by relating them to non-image variables. To do so, we employ RSD, an algorithm for relational subgroup discovery, with the cluster membership of patients as target variable. Our results enable interesting interpretations of differences in brain metabolism in terms of demographic and clinical variables. The approach was implemented and tested on an exceptionally large data collection of patients with different types of dementia. It comprises 10 GB of image data from 454 PET scans, and 42 variables from psychological and demographical data organized in 11 relations of a relational database. We believe that explaining medical images in terms of other variables (patient records, demographic information, etc.) is a challenging new and rewarding area for data mining research.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Interpreting PET Scans by Structured Patient Data: A Data Mining Case Study in Dementia Research

Hapfelmeier¹,

Schmidt²,

Mueller³

et al. 2008

2008 Eighth IEEE International Conference on Data Mining

View full text Add to dashboard Cite

show abstract

“…In order to evaluate how BCP performs, it will be compared with RSD (Železný and Lavrač, 2006), a well-known propositionalization algorithm for which an implementation is available at (http://labe.felk.cvut.cz/~zelezny/rsd). RSD is a system which tackles the Relational Subgroup Discovery problem: given a population of individuals and a property of interest, RSD seeks to find population subgroups that are as large as possible and have the most unusual distribution characteristics.…”

Section: Propositionalizationmentioning

confidence: 99%

“…We have compared CILP++ with Aleph (Srinivasan, 2007) -a state-of-the-art ILP system based on Progol -and compared BCP with a well-known propositionalization method, RSD (Železný and Lavrač, 2006), using neural networks and the C4.5 decision tree learner (Quinlan, 1993), on a number of benchmarks: four Alzheimer's datasets (King and Srinivasan, 1995) and the Mutagenesis (Srinivasan and Muggleton, 1994), KRK (Bain and Muggleton, 1994) and UW-CSE (Richardson and Domingos, 2006) datasets. Several aspects were empirically evaluated: standard classification accuracy using cross-validation and runtime measurements, how BCP performs in comparison with RSD, and how CILP++ performs in different settings using feature selection (Guyon and Elisseeff, 2003).…”

Section: Introductionmentioning

confidence: 99%

Fast relational learning using bottom clause propositionalization with artificial neural networks

2013

View full text Add to dashboard Cite

Relational learning can be described as the task of learning first-order logic rules from examples. It has enabled a number of new machine learning applications, e.g. graph mining and link analysis. Inductive Logic Programming (ILP) performs relational learning either directly by manipulating first-order rules or through propositionalization, which translates the relational task into an attribute-value learning task by representing subsets of relations as features. In this paper, we introduce a fast method and system for relational learning based on a novel propositionalization called Bottom Clause Propositionalization (BCP). Bottom clauses are boundaries in the hypothesis search space used by ILP systems Progol and Aleph. Bottom clauses carry semantic meaning and can be mapped directly onto numerical vectors, simplifying the feature extraction process. We have integrated BCP with a well-known neural-symbolic system, C-IL 2 P, to perform learning from numerical vectors. C-IL 2 P uses background knowledge in the form of propositional logic programs to build a neural network. The integrated system, which we call CILP++, handles first-order logic knowledge and is available for download from Sourceforge. We have evaluated CILP++ on seven ILP datasets, comparing results with Aleph and a well-known propositionalization method, RSD. The results show that CILP++ can achieve accuracy comparable to Aleph, while being generally faster, BCP achieved statistically significant improvement in accuracy in comparison with RSD when running with a neural network, but BCP and RSD perform similarly when running with C4.5. We have also extended CILP++ to include a statistical feature selection method, mRMR, with preliminary results indicating that a reduction of more than 90% of features can be achieved with a small loss of accuracy.

show abstract

“…Following a significant body of work in ILP [1], in our work a feature corresponds to a clause, and it holds for a sequence if the clause satisfies the sequence. We followed the approach described in [2] to map each sequence in a set of features.…”

Section: Clustering Protein Sequencesmentioning

confidence: 99%

Partitional Clustering of Protein Sequences – An Inductive Logic Programming Approach

Fonseca

Costa

Camacho

et al. 2009

Distributed Computing, Artificial Intelligence, Bioinformatics, Soft Computing, and Ambient Assisted Living

View full text Add to dashboard Cite

show abstract

Propositionalization-based relational subgroup discovery with RSD

Cited by 87 publications

References 29 publications

Interpreting PET Scans by Structured Patient Data: A Data Mining Case Study in Dementia Research

Interpreting PET Scans by Structured Patient Data: A Data Mining Case Study in Dementia Research

Fast relational learning using bottom clause propositionalization with artificial neural networks

Partitional Clustering of Protein Sequences – An Inductive Logic Programming Approach

Contact Info

Product

Resources

About