Matches in SemOpenAlex for { <https://semopenalex.org/work/W191146219> ?p ?o ?g. }
Showing items 1 to 69 of
69
with 100 items per page.
- W191146219 abstract "We introduce a graph-based relational concept learning approach and its implementation in the SubdueCL system. We describe three approaches of learning systems that use a graph representation: the Galois Lattice, Conceptual Graphs, and Subdue. The study of these systems gave us significant insight on how the concept-learning task is conducted in a graph-based system and also on how we can elaborate a theoretical analysis of our system. In the theory part we performed a PAC learning analysis of our system where we found that it is possible to PAC learn single-graphs with our approach using a polynomial number of examples. For the case of learning multiple graphs in DNF, the number of examples needed is exponential, but with experimental results we show that SubdueCL does not need that many examples to learn. We also compared our graph-based concept learning approach to the Galois lattice formal framework, which attempts to constrain the search space to polynomial size. We show that SubdueCL also creates a Galois lattice except that SubdueCL's lattice may add graphs which are super-graphs of others while removing graphs that are not connected. Then, we present SubdueCL, our graph-based concept learner, and show some experimental results including a comparison of SubdueCL with the Inductive Logic Programming systems FOIL and Progol. Results show that SubdueCL is competitive with logic-based learners using structural data. Results in an artificial domain show that SubdueCL learns successfully in training example graphs of various size, density and noise. In the case of noise (when the positive concept also exists in the negative graphs), SubdueCL learns several more specific graphs found in the positive graphs, but not in the negative graphs. In the case of density, SubdueCL tends to learn more sub-graphs (to represent the concept) as the graph density increases. Results from the theoretical and empirical analyses indicate that our graph-based approach to concept learning is efficient and effective." @default.
- W191146219 created "2016-06-24" @default.
- W191146219 creator A5044712355 @default.
- W191146219 creator A5055897947 @default.
- W191146219 date "2001-01-01" @default.
- W191146219 modified "2023-09-23" @default.
- W191146219 title "Empirical and theoretical analysis of relational concept learning using a graph-based representation" @default.
- W191146219 hasPublicationYear "2001" @default.
- W191146219 type Work @default.
- W191146219 sameAs 191146219 @default.
- W191146219 citedByCount "3" @default.
- W191146219 crossrefType "journal-article" @default.
- W191146219 hasAuthorship W191146219A5044712355 @default.
- W191146219 hasAuthorship W191146219A5055897947 @default.
- W191146219 hasConcept C118615104 @default.
- W191146219 hasConcept C121332964 @default.
- W191146219 hasConcept C124101348 @default.
- W191146219 hasConcept C154945302 @default.
- W191146219 hasConcept C161301231 @default.
- W191146219 hasConcept C177877439 @default.
- W191146219 hasConcept C234837 @default.
- W191146219 hasConcept C24890656 @default.
- W191146219 hasConcept C2779382394 @default.
- W191146219 hasConcept C2781204021 @default.
- W191146219 hasConcept C33923547 @default.
- W191146219 hasConcept C41008148 @default.
- W191146219 hasConcept C5655090 @default.
- W191146219 hasConcept C80444323 @default.
- W191146219 hasConceptScore W191146219C118615104 @default.
- W191146219 hasConceptScore W191146219C121332964 @default.
- W191146219 hasConceptScore W191146219C124101348 @default.
- W191146219 hasConceptScore W191146219C154945302 @default.
- W191146219 hasConceptScore W191146219C161301231 @default.
- W191146219 hasConceptScore W191146219C177877439 @default.
- W191146219 hasConceptScore W191146219C234837 @default.
- W191146219 hasConceptScore W191146219C24890656 @default.
- W191146219 hasConceptScore W191146219C2779382394 @default.
- W191146219 hasConceptScore W191146219C2781204021 @default.
- W191146219 hasConceptScore W191146219C33923547 @default.
- W191146219 hasConceptScore W191146219C41008148 @default.
- W191146219 hasConceptScore W191146219C5655090 @default.
- W191146219 hasConceptScore W191146219C80444323 @default.
- W191146219 hasLocation W1911462191 @default.
- W191146219 hasOpenAccess W191146219 @default.
- W191146219 hasPrimaryLocation W1911462191 @default.
- W191146219 hasRelatedWork W1553699479 @default.
- W191146219 hasRelatedWork W1993793464 @default.
- W191146219 hasRelatedWork W2071744657 @default.
- W191146219 hasRelatedWork W2124996875 @default.
- W191146219 hasRelatedWork W2139031914 @default.
- W191146219 hasRelatedWork W2139906443 @default.
- W191146219 hasRelatedWork W2508478068 @default.
- W191146219 hasRelatedWork W2550842820 @default.
- W191146219 hasRelatedWork W3006129586 @default.
- W191146219 hasRelatedWork W3010642661 @default.
- W191146219 hasRelatedWork W3015255358 @default.
- W191146219 hasRelatedWork W3023810035 @default.
- W191146219 hasRelatedWork W3036653589 @default.
- W191146219 hasRelatedWork W3038274943 @default.
- W191146219 hasRelatedWork W3103412713 @default.
- W191146219 hasRelatedWork W3104211877 @default.
- W191146219 hasRelatedWork W3121197195 @default.
- W191146219 hasRelatedWork W3126412650 @default.
- W191146219 hasRelatedWork W3205454756 @default.
- W191146219 hasRelatedWork W587388447 @default.
- W191146219 isParatext "false" @default.
- W191146219 isRetracted "false" @default.
- W191146219 magId "191146219" @default.
- W191146219 workType "article" @default.