Deep transfer via second-order Markov logic

ABSTRACT

Standard inductive learning requires that training and test instances come from the same distribution. Transfer learning seeks to remove this restriction. In shallow transfer, test instances are from the same domain, but have a different distribution. In deep transfer, test instances are from a different domain entirely (i.e., described by different predicates). Humans routinely perform deep transfer, but few learning systems, if any, are capable of it. In this paper we propose an approach based on a form of second-order Markov logic. Our algorithm discovers structural regularities in the source domain in the form of Markov logic formulas with predicate variables, and instantiates these formulas with predicates from the target domain. Using this approach, we have successfully transferred learned knowledge among molecular biology, social network and Web domains. The discovered patterns include broadly useful properties of predicates, like symmetry and transitivity, and relations among predicates, such as various forms of homophily.

REFERENCES

• 1: Banerjee, B., Liu, Y., & Youngblood, G. (Eds.). (2006). Workshop on structural knowledge transfer for machine learning.
• 2: Baxter, J., Caruana, R., Mitchell, T., Pratt, L. Y., Silver, D. L., & Thrun, S. (Eds.). (1995). Workshop on learning to learn: Knowledge consolidation and transfer in inductive systems.
• 3: Bridewell, W., & Todorovski, L. (2007). Learning declarative bias. Proc. 17th Conf. on Inductive Logic Programming (pp. 63--77).
• 4: Mark Craven , Seán Slattery, Relational learning with statistical predicate invention: better models for hypertext, Machine Learning, v.43 n.1/2, p.97-119, April/May 2001 [doi>10.1023/A:1007676901476]
• 5: Davis, J., Burnside, E., Dutra, I., Page, D., & Costa, V. S. (2005). An integrated approach to learning Bayesian networks of rules. Proc. 16th Eur. Conf. on Machine Learning (pp. 84--95).
• 6: Luc De Raedt , Maurice Bruynooghe, Interactive Concept-Learning and Constructive Induction by Analogy, Machine Learning, v.8 n.2, p.107-150, March 1992 [doi>10.1023/A:1022664419589]
• 7: Stephen Della Pietra , Vincent Della Pietra , John Lafferty, Inducing Features of Random Fields, IEEE Transactions on Pattern Analysis and Machine Intelligence, v.19 n.4, p.380-393, April 1997 [doi>10.1109/34.588021]
• 8: Nada Lavrac , Saso Dzeroski, Relational Data Mining, Springer-Verlag New York, Inc., Secaucus, NJ, 2001
• 9: Brian Falkenhainer , Kenneth D. Forbus , Dedre Gentner, The structure-mapping engine: algorithm and examples, Artificial Intelligence, v.41 n.1, p.1-63, Nov. 1989 [doi>10.1016/0004-3702(89)90077-5]
• 10: Stanley Kok , Pedro Domingos, Learning the structure of Markov logic networks, Proceedings of the 22nd international conference on Machine learning, p.441-448, August 07-11, 2005, Bonn, Germany [doi>10.1145/1102351.1102407]
• 11: Stanley Kok , Pedro Domingos, Statistical predicate invention, Proceedings of the 24th international conference on Machine learning, p.433-440, June 20-24, 2007, Corvalis, Oregon [doi>10.1145/1273496.1273551]
• 12: Kok, S., Sumner, M., Richardson, M., Singla, P., Poon, H., Lowd, D., Wang, J., & Domingos, P. (2009). The Alchemy system for statistical relational AI (Technical Report). Department of Computer Science and Engineering, University of Washington, Seattle, WA. http://alchemy.cs.washington.edu.
• 13: Daniel Lowd , Pedro Domingos, Efficient Weight Learning for Markov Logic Networks, Proceedings of the 11th European conference on Principles and Practice of Knowledge Discovery in Databases, September 17-21, 2007, Warsaw, Poland [doi>10.1007/978-3-540-74976-9_21]
• 14: Mewes, H. W., Frishman, D., Gruber, C., Geier, B., Haase, D., Kaps, A., Lemcke, K., Mannhaupt, G., Pfeiffer, F., Schüller, C., Stocker, S., & Weil, B. (2000). MIPS: a database for genomes and protein sequences. Nucleic Acids Research, 28, 37--40.
• 15: Mihalkova, L., Huynh, T., & Mooney, R. J. (2007). Mapping and revising Markov logic networks for transfer learning. Proc. 22nd Conf. on Artificial Intelligence (pp. 608--614).
• 16: Silver, D., Bakir, G., Bennett, K., Caruana, R., Pontil, M., Russell, S., & Tadepalli, P. (Eds.). (2005). Workshop on inductive transfer: 10 years later.
• 17: Silverstein, G., & Pazzani, M. J. (1991). Relational clichéés: Constraining constructive induction during relational learning. Proc. 8th Intl. Conf. on Machine Learning (pp. 203--207).
• 18: Taskar, B., Abbeel, P., & Koller, D. (2002). Discriminative probabilistic models for relational data. Proc. 18th Conf. on Uncertainty in Artificial Intelligence (pp. 485--492).
• 19: Taylor, M., Fern, A., & Driessens, K. (Eds.). (2008). Workshop on transfer learning for complex tasks.

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
      I.2.6 Learning
          Subjects: Induction

Additional Classification:
  G. Mathematics of Computing
  G.3 PROBABILITY AND STATISTICS
      Subjects: Markov processes

  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
      I.2.4 Knowledge Representation Formalisms and Methods
          Subjects: Predicate logic
  I.5 PATTERN RECOGNITION
      I.5.2 Design Methodology
  I.6 SIMULATION AND MODELING
      I.6.5 Model Development
          Subjects: Modeling methodologies

General Terms:
Algorithms, Design, Theory