structured probability spaces
play

Structured Probability Spaces Guy Van den Broeck Southern - PowerPoint PPT Presentation

Tractable Learning in Structured Probability Spaces Guy Van den Broeck Southern California Machine Learning Symposium Nov 18, 2016 Structured probability spaces? Running Example Courses: Data Logic (L) Knowledge Representation (K)


  1. Tractable Learning in Structured Probability Spaces Guy Van den Broeck Southern California Machine Learning Symposium Nov 18, 2016

  2. Structured probability spaces?

  3. Running Example Courses: Data • Logic (L) • Knowledge Representation (K) • Probability (P) • Artificial Intelligence (A) Constraints • Must take at least one of Probability or Logic. • Probability is a prerequisite for AI. • The prerequisites for KR is either AI or Logic.

  4. Probability Space unstructured L K P A 0 0 0 0 0 0 0 1 0 0 1 0 0 0 1 1 0 1 0 0 0 1 0 1 0 1 1 0 0 1 1 1 1 0 0 0 1 0 0 1 1 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 1 0 1 1 1 1

  5. Structured Probability Space unstructured structured L K P A L K P A 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 • Must take at least one of 0 0 1 0 0 0 1 0 Probability or Logic. 0 0 1 1 0 0 1 1 • Probability is a prerequisite for AI. 0 1 0 0 0 1 0 0 • 0 1 0 1 The prerequisites for KR is 0 1 0 1 0 1 1 0 either AI or Logic. 0 1 1 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 0 0 1 0 0 1 1 0 0 1 1 0 1 0 1 0 1 0 7 out of 16 instantiations 1 0 1 1 1 0 1 1 are impossible 1 1 0 0 1 1 0 0 1 1 0 1 1 1 0 1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1

  6. Learning with Constraints Learn a statistical model that assigns zero probability to instantiations that violate the constraints.

  7. Example: Video [Lu, W. L., Ting, J. A., Little, J. J., & Murphy, K. P. (2013). Learning to track and identify players from broadcast sports videos.]

  8. Example: Video [Lu, W. L., Ting, J. A., Little, J. J., & Murphy, K. P. (2013). Learning to track and identify players from broadcast sports videos.]

  9. Example: Language • Non-local dependencies: At least one verb in each sentence [Chang, M., Ratinov, L., & Roth, D. (2008). Constraints as prior knowledge],…, [ Chang, M. W., Ratinov, L., & Roth, D. (2012). Structured learning with constrained conditional models.], [https://en.wikipedia.org/wiki/Constrained_conditional_model]

  10. Example: Language • Non-local dependencies: At least one verb in each sentence • Sentence compression If a modifier is kept, its subject is also kept [Chang, M., Ratinov, L., & Roth, D. (2008). Constraints as prior knowledge],…, [ Chang, M. W., Ratinov, L., & Roth, D. (2012). Structured learning with constrained conditional models.], [https://en.wikipedia.org/wiki/Constrained_conditional_model]

  11. Example: Language • Non-local dependencies: At least one verb in each sentence • Sentence compression If a modifier is kept, its subject is also kept • Information extraction [Chang, M., Ratinov, L., & Roth, D. (2008). Constraints as prior knowledge],…, [ Chang, M. W., Ratinov, L., & Roth, D. (2012). Structured learning with constrained conditional models.], [https://en.wikipedia.org/wiki/Constrained_conditional_model]

  12. Example: Language • Non-local dependencies: At least one verb in each sentence • Sentence compression If a modifier is kept, its subject is also kept • Information extraction Semantic role labeling • … and many more! [Chang, M., Ratinov, L., & Roth, D. (2008). Constraints as prior knowledge],…, [ Chang, M. W., Ratinov, L., & Roth, D. (2012). Structured learning with constrained conditional models.], [https://en.wikipedia.org/wiki/Constrained_conditional_model]

  13. Example: Deep Learning [Graves, A., Wayne, G., Reynolds, M., Harley, T., Danihelka, I., Grabska- Barwińska , A., et al.. (2016). Hybrid computing using a neural network with dynamic external memory. Nature , 538 (7626), 471-476.]

  14. Example: Deep Learning [Graves, A., Wayne, G., Reynolds, M., Harley, T., Danihelka, I., Grabska- Barwińska , A., et al.. (2016). Hybrid computing using a neural network with dynamic external memory. Nature , 538 (7626), 471-476.]

  15. Example: Deep Learning [Graves, A., Wayne, G., Reynolds, M., Harley, T., Danihelka, I., Grabska- Barwińska , A., et al.. (2016). Hybrid computing using a neural network with dynamic external memory. Nature , 538 (7626), 471-476.]

  16. Example: Deep Learning [Graves, A., Wayne, G., Reynolds, M., Harley, T., Danihelka, I., Grabska- Barwińska , A., et al.. (2016). Hybrid computing using a neural network with dynamic external memory. Nature , 538 (7626), 471-476.]

  17. What are people doing now? • Ignore • Hack your way around • Handcraft into models • Use specialized distributions • Find non-structured encoding • Try to learn constraints

  18. What are people doing now? • Ignore • Hack your way around • Handcraft into models Accuracy ? • Use specialized distributions Specialized skill ? • Find non-structured encoding Impossible ? • Try to learn constraints Intractable inference ? Intractable learning ? Waste parameters ? Risk predicting out of space ? + you are on your own 

  19. Structured Probability Spaces • Everywhere in ML! – Configuration problems, video, text, deep learning – Planning and diagnosis (physics) – Cooking scenarios (interpreting videos) – Combinatorial objects: parse trees, rankings, directed acyclic graphs, trees, simple paths, game traces, etc.

  20. Structured Probability Spaces • Everywhere in ML! – Configuration problems, video, text, deep learning – Planning and diagnosis (physics) – Cooking scenarios (interpreting videos) – Combinatorial objects: parse trees, rankings, directed acyclic graphs, trees, simple paths, game traces, etc. • Representations: constrained conditional models, mixed networks, probabilistic logics.

  21. Structured Probability Spaces • Everywhere in ML! – Configuration problems, video, text, deep learning – Planning and diagnosis (physics) – Cooking scenarios (interpreting videos) – Combinatorial objects: parse trees, rankings, directed acyclic graphs, trees, simple paths, game traces, etc. • Representations: constrained conditional models, mixed networks, probabilistic logics. No ML boxes out there that take constraints as input! 

  22. The Problem / The ML Box Goal: Constraints as important as data! General purpose! Data Probabilistic Model Learning (Distribution) Constraints

  23. Specification Language: Logic

  24. Structured Probability Space unstructured structured L K P A L K P A 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 • Must take at least one of 0 0 1 0 0 0 1 0 Probability or Logic. 0 0 1 1 0 0 1 1 • Probability is a prerequisite for AI. 0 1 0 0 0 1 0 0 • 0 1 0 1 The prerequisites for KR is 0 1 0 1 0 1 1 0 either AI or Logic. 0 1 1 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 0 0 1 0 0 1 1 0 0 1 1 0 1 0 1 0 1 0 7 out of 16 instantiations 1 0 1 1 1 0 1 1 are impossible 1 1 0 0 1 1 0 0 1 1 0 1 1 1 0 1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1

  25. Boolean Constraints unstructured structured L K P A L K P A 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 1 0 0 0 1 0 0 0 1 1 0 0 1 1 0 1 0 0 0 1 0 0 0 1 0 1 0 1 0 1 0 1 1 0 0 1 1 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 0 0 1 0 0 1 1 0 0 1 1 0 1 0 1 0 1 0 7 out of 16 instantiations 1 0 1 1 1 0 1 1 are impossible 1 1 0 0 1 1 0 0 1 1 0 1 1 1 0 1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1

  26. Combinatorial Objects: Rankings rank sushi rank sushi 1 fatty tuna 1 shrimp 10 items : 2 sea urchin 2 sea urchin 3,628,800 3 salmon roe 3 salmon roe rankings 4 shrimp 4 fatty tuna 5 tuna 5 tuna 6 squid 6 squid 20 items : 7 tuna roll 7 tuna roll 2,432,902,008,176,640,000 8 see eel 8 see eel rankings 9 egg 9 egg 10 cucumber roll 10 cucumber roll

  27. Combinatorial Objects: Rankings rank sushi rank sushi A ij item i at position j 1 fatty tuna 1 shrimp (n items require n 2 2 sea urchin 2 sea urchin Boolean variables) 3 salmon roe 3 salmon roe 4 shrimp 4 fatty tuna 5 tuna 5 tuna 6 squid 6 squid 7 tuna roll 7 tuna roll 8 see eel 8 see eel 9 egg 9 egg 10 cucumber roll 10 cucumber roll

  28. Combinatorial Objects: Rankings rank sushi rank sushi A ij item i at position j 1 fatty tuna 1 shrimp (n items require n 2 2 sea urchin 2 sea urchin Boolean variables) 3 salmon roe 3 salmon roe 4 shrimp 4 fatty tuna An item may be assigned 5 tuna 5 tuna to more than one position 6 squid 6 squid 7 tuna roll 7 tuna roll A position may contain 8 see eel 8 see eel more than one item 9 egg 9 egg 10 cucumber roll 10 cucumber roll

  29. Encoding Rankings in Logic A ij : item i at position j pos 1 pos 2 pos 3 pos 4 item 1 A 11 A 12 A 13 A 14 item 2 A 21 A 22 A 23 A 24 item 3 A 31 A 32 A 33 A 34 item 4 A 41 A 42 A 43 A 44

  30. Encoding Rankings in Logic A ij : item i at position j constraint: each item i assigned to a unique position ( n constraints) pos 1 pos 2 pos 3 pos 4 item 1 A 11 A 12 A 13 A 14 item 2 A 21 A 22 A 23 A 24 item 3 A 31 A 32 A 33 A 34 item 4 A 41 A 42 A 43 A 44

  31. Encoding Rankings in Logic A ij : item i at position j constraint: each item i assigned to a unique position ( n constraints) pos 1 pos 2 pos 3 pos 4 item 1 A 11 A 12 A 13 A 14 item 2 A 21 A 22 A 23 A 24 constraint: each position j assigned item 3 A 31 A 32 A 33 A 34 a unique item ( n constraints) item 4 A 41 A 42 A 43 A 44

Recommend


More recommend