Learning Distributed Representations of Relational Data using Linear Relational Embedding
Linear Relational Embedding (LRE) is a new method of learning a distributed representation of concepts from data consisting of binary relations between concepts. The final goal of LRE is to be able to generalize, i.e. to infer new relations among the concepts. The version presented here is capable of handling incomplete information and multiple correct answers. We present results on two simple domains, that show an excellent generalization performance.
Unable to display preview. Download preview PDF.
- Geoffrey E. Hinton. Learning distributed representations of concepts. In Proceedings of the Eighth Annual Conference of the Cognitive Science Society, pages 1–12. Erlbaum, NJ, 1986.Google Scholar
- Thomas K. Landauer, Darrel Laham, and Peter Foltz. Learning human-like knowledge by singular value decomposition: A progress report. In Michael I. Jordan, Michael J. Kearns, and sara A. Solla, editors, Advances in Neural Processing Information Systems 10, pages 45–51. The MIT Press, Cambridge Massachusetts, 1998.Google Scholar
- Alberto Paccanaro and Geoffrey E. Hinton. Extracting distributed representations of concepts and relations from positive and negative propositions. In Proceedings of the International Joint Conference on Neural Networks, IJCNN 2000. 2000.Google Scholar
- Alberto Paccanaro and Geoffrey E. Hinton. Learning distributed representations by mapping concepts and relations into a linear space. In Pat Langley, editor, Proceedings of the Seventeenth International Conference on Machine Learning, ICML2000, pages 711–718. Morgan Kaufmann Publishers, Stanford University, San Francisco, 2000.Google Scholar
- Alberto Paccanaro and Geoffrey E. Hinton. Learning distributed representation of concepts using linear relational embedding, to appear in IEEE Trans. on Knowledge and Data Engineering - special issue on Connectionists Models for Learning in Structured Domains, 2001.Google Scholar
- J. R. Quinlan. Learning logical definitions from relations. Machine Learning, 5:239–266, 1990.Google Scholar
- F. W. Young and R. M. Hamer. Multidimensional Scaling: History, Theory and Applications. Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers„ 1987.Google Scholar