Approximate tree kernels

Research output: Journal contributionsJournal articlesResearchpeer-review

Authors

  • Konrad Rieck
  • Tammo Krueger
  • Ulf Brefeld
  • Klaus Robert Müller

Convolution kernels for trees provide simple means for learning with tree-structured data. The computation time of tree kernels is quadratic in the size of the trees, since all pairs of nodes need to be compared. Thus, large parse trees, obtained from HTML documents or structured network data, render convolution kernels inapplicable. In this article, we propose an effective approximation technique for parse tree kernels. The approximate tree kernels (ATKs) limit kernel computation to a sparse subset of relevant subtrees and discard redundant structures, such that training and testing of kernel-based learning methods are significantly accelerated. We devise linear programming approaches for identifying such subsets for supervised and unsupervised learning tasks, respectively. Empirically, the approximate tree kernels attain run-time improvements up to three orders of magnitude while preserving the predictive accuracy of regular tree kernels. For unsupervised tasks, the approximate tree kernels even lead to more accurate predictions by identifying relevant dimensions in feature space.

Original languageEnglish
JournalJournal of Machine Learning Research
Volume11
Pages (from-to)555-580
Number of pages26
ISSN1532-4435
Publication statusPublished - 02.2010
Externally publishedYes

Documents

Links

Recently viewed

Publications

  1. A framework for business model development in technology-driven start-ups
  2. Evaluation of standard ERP software implementation approaches in terms of their capability for business process optimization
  3. Conceptions of problem solving mathematics teaching
  4. What´s in a net? or: The end of the average
  5. Technological System and the Problem of Desymbolization
  6. Web-scale extension of RDF knowledge bases from templated websites
  7. Errors in Working with Office Computers
  8. Understanding the modes of use and availability of critical metals-An expert-based scenario analysis for the case of indium
  9. Learning from partially annotated sequences
  10. Other spaces
  11. Machine Learning Applications
  12. The Creation of the Concept through the Interaction of Philosophy with Science and Art
  13. Do guided internet-based interventions result in clinically relevant changes for patients with depression?
  14. Networking for the environment
  15. Design of an Information-Based Distributed Production Planning System
  16. Teaching methods for modelling problems and students’ task-specific enjoyment, value, interest and self-efficacy expectations
  17. Topic selection and development in learner-native speaker voice-based telecollaborative discourse
  18. Adaptive control of the nonlinear dynamic behavior of the cantilever-sample system of an atomic force microscope
  19. Estimation and interpretation of a Heckman selection model with endogenous covariates
  20. The buffering effect of selection, optimization, and compensation strategy use on the relationship between problem solving demands and occupational well-being
  21. Holistic and scalable ranking of RDF data
  22. Polar Coordinates and Interactive Learning
  23. Effects Of Different Order Processing Strategies On Operating Curves Of Logistic Models
  24. Meta-Image – a collaborative environment for the image discourse
  25. Recontextualizing Anthropomorphic Metaphors in Organization Studies
  26. Switching Dispatching Rules with Gaussian Processes
  27. Global fern and lycophyte richness explained: How regional and local factors shape plot richness
  28. Clause identification using entropy guided transformation learning
  29. ℓp-norm multiple kernel learning
  30. Individual Differences in Infants' Speech Segmentation Performance

Press / Media

  1. Wieder gefragt