Tuesday, September 02, 2008

Machine Learning: Ten Challenges for the Next Ten Years

In the October 2008 special issue on inductive logic programming in Machine
Learning
, Dietterich et al [1] lay-out the following ten outstanding problems for the next ten years:
  1. Statistical predicate invention
  2. Generalizing across domains
  3. Learning many levels of structure
  4. Deep combination of learning and inference
  5. Learning to map between representations
  6. Learning in the large
  7. Structured prediction with intractable inference
  8. Reinforcement learning with structured time
  9. Expanding SRL (Statistical Relational Learning) to statistical relational AI
  10. Learning to debug programs


[1] Thomas G. Dietterich, Pedro Domingos, Lise Getoor, Stephen Muggleton, Prasad Tadepalli (2008). Structured machine learning: the next ten years Machine Learning, 73 (1), 3-23 DOI: 10.1007/s10994-008-5079-1

No comments: