causal reasoning and inference with causal bayes nets
play

Causal reasoning and inference with causal Bayes nets Alexander - PowerPoint PPT Presentation

May 14, 2023 •26 likes •1.09k views

Causal reasoning and inference with causal Bayes nets Alexander Gebharter Duesseldorf Center for Logic and Philosophy of Science Heinrich Heine University Duesseldorf 28.04.2016 Alexander Gebharter (DCLPS) Reasoning & inference with CBNs

  1. Causal Bayes nets Definition ( d -connection/ d -separation) X and Y are d -connected by Z ⊆ V \{ X , Y } if and only if X and Y are connected by a causal path π such that (i) no non-collider on π is in Z , and (ii) every collider on π is in Z or has an effect in Z . X and Y are d -separated by Z iff they are not d -connected by Z . Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 7 / 26

  2. Causal Bayes nets Definition ( d -connection/ d -separation) X and Y are d -connected by Z ⊆ V \{ X , Y } if and only if X and Y are connected by a causal path π such that (i) no non-collider on π is in Z , and (ii) every collider on π is in Z or has an effect in Z . X and Y are d -separated by Z iff they are not d -connected by Z . Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 7 / 26

  3. Causal Bayes nets Definition ( d -connection/ d -separation) X and Y are d -connected by Z ⊆ V \{ X , Y } if and only if X and Y are connected by a causal path π such that (i) no non-collider on π is in Z , and (ii) every collider on π is in Z or has an effect in Z . X and Y are d -separated by Z iff they are not d -connected by Z . Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 7 / 26

  4. Causal Bayes nets Definition ( d -connection/ d -separation) X and Y are d -connected by Z ⊆ V \{ X , Y } if and only if X and Y are connected by a causal path π such that (i) no non-collider on π is in Z , and (ii) every collider on π is in Z or has an effect in Z . X and Y are d -separated by Z iff they are not d -connected by Z . Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 7 / 26

  5. Causal Bayes nets Definition ( d -connection condition) A causal model satisfies the d -connection condition if and only if for all X , Y ∈ V and Z ⊆ V \{ X , Y } : If Dep ( X , Y | Z ) , then X and Y are d -connected by Z . Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 8 / 26

  6. Causal Bayes nets Definition (causal Markov condition) A causal model satisfies the causal Markov condition (CMC) if and only if every X is probabilistically independent of its non-effects conditional on its direct causes. (cf. Spirtes et al., 2000, p. 29) CMC determines the following Markov factorization: n � P ( x 1 , ..., x n ) = P ( x i | par ( X i )) (1) i = 1 Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 9 / 26

  7. Causal Bayes nets Definition (causal Markov condition) A causal model satisfies the causal Markov condition (CMC) if and only if every X is probabilistically independent of its non-effects conditional on its direct causes. (cf. Spirtes et al., 2000, p. 29) CMC determines the following Markov factorization: n � P ( x 1 , ..., x n ) = P ( x i | par ( X i )) (1) i = 1 Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 9 / 26

  8. Causal Bayes nets P ( a , b , c , d , e ) = P ( a ) · P ( b | a ) · P ( c | a ) · P ( d | b , c ) · P ( e | d ) Indep ( B , C | A ) Indep ( C , B | A ) Indep ( D , A |{ B , C } ) Indep ( E , { A , B , C }| D ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 10 / 26

  9. Causal Bayes nets P ( a , b , c , d , e ) = P ( a ) · P ( b | a ) · P ( c | a ) · P ( d | b , c ) · P ( e | d ) Indep ( B , C | A ) Indep ( C , B | A ) Indep ( D , A |{ B , C } ) Indep ( E , { A , B , C }| D ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 10 / 26

  10. Causal Bayes nets The causal Markov condition is assumed to be satisfied by causal models that satisfy the causal sufficiency condition. Definition (causal sufficiency condition) A causal model satisfies the causal sufficiency condition if and only if every common cause C of every pair X , Y ∈ V is in V or is fixed to a certain value c . Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 11 / 26

  11. Causal Bayes nets A causal model that satisfies CMC satisfies the causal faithfulness condition (CFC) if and only if the independencies implied by CMC are all the independencies in the model (cf. Spirtes et al., 2000, p. 31). Generalized: Definition (causal faithfulness condition) A causal model satisfies the causal faithfulness condition if and only if every d -connection implies a probabilistic dependence. (cf. Schurz & Gebharter, 2015, sec. 3.2) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 12 / 26

  12. Causal Bayes nets A causal model that satisfies CMC satisfies the causal faithfulness condition (CFC) if and only if the independencies implied by CMC are all the independencies in the model (cf. Spirtes et al., 2000, p. 31). Generalized: Definition (causal faithfulness condition) A causal model satisfies the causal faithfulness condition if and only if every d -connection implies a probabilistic dependence. (cf. Schurz & Gebharter, 2015, sec. 3.2) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 12 / 26

  13. Causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 13 / 26

  14. Causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 13 / 26

  15. Causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 13 / 26

  16. Causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 13 / 26

  17. Causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 13 / 26

  18. Causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 13 / 26

  19. Causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 13 / 26

  20. Causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 13 / 26

  21. Causal Bayes nets Introduction Causal Bayes nets Intervention and observation Causal reasoning with causal Bayes nets Causal discovery with causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 14 / 26

  22. Causal Bayes nets Introduction Causal Bayes nets Intervention and observation Causal reasoning with causal Bayes nets Causal discovery with causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 14 / 26

  23. Intervention and observation CBNs allow for distinguishing intervention from observation (cf. Pearl, 2009, sec. 1.3.1; Spirtes et al., 2000, sec. 3.7.2). Sprinkler Season Wet Slippery Rain Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 15 / 26

  24. Intervention and observation CBNs allow for distinguishing intervention from observation (cf. Pearl, 2009, sec. 1.3.1; Spirtes et al., 2000, sec. 3.7.2). Sprinkler = on Season Wet Slippery Rain Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 15 / 26

  25. Intervention and observation CBNs allow for distinguishing intervention from observation (cf. Pearl, 2009, sec. 1.3.1; Spirtes et al., 2000, sec. 3.7.2). Sprinkler = on Season Wet Slippery Rain Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 15 / 26

  26. Intervention and observation Introduction Causal Bayes nets Intervention and observation Causal reasoning with causal Bayes nets Causal discovery with causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 16 / 26

  27. Intervention and observation Introduction Causal Bayes nets Intervention and observation Causal reasoning with causal Bayes nets Causal discovery with causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 16 / 26

  28. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation: P ( sl 1 | sp on ) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 17 / 26

  29. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation: P ( sl 1 | sp on ) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 17 / 26

  30. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation: P ( sl 1 | sp on ) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 17 / 26

  31. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation: P ( sl 1 | sp on ) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 17 / 26

  32. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation: P ( sl 1 | sp on ) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 17 / 26

  33. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation: P ( sl 1 | sp on ) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 17 / 26

  34. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation: P ( sl 1 | sp on ) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 17 / 26

  35. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation: P ( sl 1 | sp on ) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 17 / 26

  36. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation: P ( sl 1 | sp on ) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( sp on | se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 17 / 26

  37. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation generalized: � u P ( y , x , u ) P ( y | x ) = w P ( x , w ) , where U = V \{ X , Y } and W = V \{ X } � Note: X and Y can also be sets of variables! Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 18 / 26

  38. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation generalized: � u P ( y , x , u ) P ( y | x ) = w P ( x , w ) , where U = V \{ X , Y } and W = V \{ X } � Note: X and Y can also be sets of variables! Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 18 / 26

  39. Causal reasoning with causal Bayes nets Sprinkler Season Wet Slippery Rain Observation generalized: � u P ( y , x , u ) P ( y | x ) = w P ( x , w ) , where U = V \{ X , Y } and W = V \{ X } � Note: X and Y can also be sets of variables! Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 18 / 26

  40. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention: P ( sl 1 | do ( sp on )) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 19 / 26

  41. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention: P ( sl 1 | do ( sp on )) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 19 / 26

  42. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention: P ( sl 1 | do ( sp on )) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 19 / 26

  43. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention: P ( sl 1 | do ( sp on )) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 19 / 26

  44. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention: P ( sl 1 | do ( sp on )) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 19 / 26

  45. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention: P ( sl 1 | do ( sp on )) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 19 / 26

  46. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention: P ( sl 1 | do ( sp on )) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 19 / 26

  47. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention: P ( sl 1 | do ( sp on )) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 19 / 26

  48. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention: P ( sl 1 | do ( sp on )) = P ( sl 1 , sp on ) P ( sp on ) � P ( sl , sp on ) = P ( sl 1 , sp on , u ) , where U = V \{ Sl , Sp } u � � P ( sl 1 , sp on , u ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we � P ( sp on , w ) , where W = V \{ Sp } P ( sp on ) = w � � P ( sp on , w ) = P ( se ) · P ( ra | se ) · P ( we | sp on , ra ) · P ( sl 1 | we ) u se , ra , we , sl Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 19 / 26

  49. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention generalized: � u P ( y , x , u ) P ( y | do ( x )) = w P ( x , w ) , where U = V \{ X , Y } and W = V \{ Y } � Note: X and Y can also be sets of variables! Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 20 / 26

  50. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention generalized: � u P ( y , x , u ) P ( y | do ( x )) = w P ( x , w ) , where U = V \{ X , Y } and W = V \{ Y } � Note: X and Y can also be sets of variables! Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 20 / 26

  51. Causal reasoning with causal Bayes nets Sprinkler = on Season Wet Slippery Rain Intervention generalized: � u P ( y , x , u ) P ( y | do ( x )) = w P ( x , w ) , where U = V \{ X , Y } and W = V \{ Y } � Note: X and Y can also be sets of variables! Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 20 / 26

  52. Intervention and observation Introduction Causal Bayes nets Intervention and observation Causal reasoning with causal Bayes nets Causal discovery with causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 21 / 26

  53. Intervention and observation Introduction Causal Bayes nets Intervention and observation Causal reasoning with causal Bayes nets Causal discovery with causal Bayes nets Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 21 / 26

  54. Causal discovery There is a multitude of search algorithms for all kinds of causal scenarios available in the literature (e.g., Spirtes et al., 2000). I will present one of these algorithms: the SGS algorithm. SGS presupposes acyclicity as well as the causal Markov condition and the faithfulness condition to hold. Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 22 / 26

  55. Causal discovery There is a multitude of search algorithms for all kinds of causal scenarios available in the literature (e.g., Spirtes et al., 2000). I will present one of these algorithms: the SGS algorithm. SGS presupposes acyclicity as well as the causal Markov condition and the faithfulness condition to hold. Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 22 / 26

  56. Causal discovery There is a multitude of search algorithms for all kinds of causal scenarios available in the literature (e.g., Spirtes et al., 2000). I will present one of these algorithms: the SGS algorithm. SGS presupposes acyclicity as well as the causal Markov condition and the faithfulness condition to hold. Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 22 / 26

  57. Causal discovery SGS algorithm (cf. Spirtes et al., 2000, p. 82) S1: Form the complete undirected graph over vertex set V . S2: Check for every X — Y for which there is a Z ⊆ V \{ X , Y } such that Indep ( X , Y | Z ) , remove the edge between X and Y . S3: For all X — Z — Y (or X − → Z — Y ) without an edge between X and Y : Orient the edges as X − → Z ← − Y iff Dep ( X , Y | M ) holds for all M ⊆ V \{ X , Y } with Z ∈ M . S4: (a) For all X − → Z — Y without an edge between X and Y : Orient Z — Y as Z − → Y . (b) If X − → ... − → Y and X — Y , then orient X — Y as X − → Y . Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 23 / 26

  58. Causal discovery SGS algorithm (cf. Spirtes et al., 2000, p. 82) S1: Form the complete undirected graph over vertex set V . S2: Check for every X — Y for which there is a Z ⊆ V \{ X , Y } such that Indep ( X , Y | Z ) , remove the edge between X and Y . S3: For all X — Z — Y (or X − → Z — Y ) without an edge between X and Y : Orient the edges as X − → Z ← − Y iff Dep ( X , Y | M ) holds for all M ⊆ V \{ X , Y } with Z ∈ M . S4: (a) For all X − → Z — Y without an edge between X and Y : Orient Z — Y as Z − → Y . (b) If X − → ... − → Y and X — Y , then orient X — Y as X − → Y . Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 23 / 26

  59. Causal discovery SGS algorithm (cf. Spirtes et al., 2000, p. 82) S1: Form the complete undirected graph over vertex set V . S2: Check for every X — Y for which there is a Z ⊆ V \{ X , Y } such that Indep ( X , Y | Z ) , remove the edge between X and Y . S3: For all X — Z — Y (or X − → Z — Y ) without an edge between X and Y : Orient the edges as X − → Z ← − Y iff Dep ( X , Y | M ) holds for all M ⊆ V \{ X , Y } with Z ∈ M . S4: (a) For all X − → Z — Y without an edge between X and Y : Orient Z — Y as Z − → Y . (b) If X − → ... − → Y and X — Y , then orient X — Y as X − → Y . Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 23 / 26

  60. Causal discovery SGS algorithm (cf. Spirtes et al., 2000, p. 82) S1: Form the complete undirected graph over vertex set V . S2: Check for every X — Y for which there is a Z ⊆ V \{ X , Y } such that Indep ( X , Y | Z ) , remove the edge between X and Y . S3: For all X — Z — Y (or X − → Z — Y ) without an edge between X and Y : Orient the edges as X − → Z ← − Y iff Dep ( X , Y | M ) holds for all M ⊆ V \{ X , Y } with Z ∈ M . S4: (a) For all X − → Z — Y without an edge between X and Y : Orient Z — Y as Z − → Y . (b) If X − → ... − → Y and X — Y , then orient X — Y as X − → Y . Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 23 / 26

  61. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  62. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  63. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  64. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  65. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  66. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  67. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  68. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  69. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  70. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  71. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  72. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  73. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  74. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

  75. Causal discovery Step 1 Step 2 Step 3 Step 4 A & B : Indep ( A , B ) A & D : Indep ( A , D | C ) Indep ( A , D |{ B , C } ) B & D : Indep ( B , D | C ) Indep ( B , D |{ A , C } ) Alexander Gebharter (DCLPS) Reasoning & inference with CBNs 28.04.2016 24 / 26

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Causal Inference An introduction based on S. Wagers course on Causal Inference (OIT 661) Imke

Causal Inference An introduction based on S. Wagers course on Causal Inference (OIT 661) Imke

Causal Inference An introduction based on S. Wagers course on Causal Inference (OIT 661) Imke Mayer November 23, 2018 Group Meeting, CMAP Outline 1. Treatment effect estimation in randomized experiments 2. Beyond a single randomized

1.48k views • 47 slides
Introduction to Causal Inference Lan Liu University of Minnesota at Twin Cities liux3771@umn.edu

Introduction to Causal Inference Lan Liu University of Minnesota at Twin Cities liux3771@umn.edu

Introduction to Causal Inference Lan Liu University of Minnesota at Twin Cities liux3771@umn.edu 1 Table of contents Causal ... or not? How Topics in Causal Inference Tools we use... Causal Inference in Industry 2 The Danger of Ice Cream

961 views • 52 slides
Bayes Nets 10-701 recitation 04-02-2013 Bayes Nets Represent dependencies between variables

Bayes Nets 10-701 recitation 04-02-2013 Bayes Nets Represent dependencies between variables

Bayes Nets 10-701 recitation 04-02-2013 Bayes Nets Represent dependencies between variables Compact representation of probability distribution Allergy Flu Encodes causal relationships Sinus Nose Headache Conditional

390 views • 22 slides
Causal Inference and Response Surface Modeling Inference and

Causal Inference and Response Surface Modeling Inference and

Causal Inference and Response Surface Modeling Inference and Representa6on DS-GA-1005 Fall 2015 Guest lecturer: Uri Shalit What is Causal Inference?

503 views • 48 slides
Modes of Statistical Inference for Causal Efgects Plus an overview of the testing based approach

Modes of Statistical Inference for Causal Efgects Plus an overview of the testing based approach

Modes of Statistical Inference for Causal Efgects Plus an overview of the testing based approach to causal inference for experiments on networks CSBS Causal Inference Workshop @ Illinois Jake Bowers Political Science & Statistics

1k views • 88 slides
{ a) policy evaluation (treatment effects ) Old b) attribution but for gems

{ a) policy evaluation (treatment effects ) Old b) attribution but for gems

JSM-2016 presented: 8.1.16 OUTLINE CAUSAL INFERENCE IN STATISTICS: 1. The causal revolution from statistics to policy intervention to counterfactuals 2. The fundamental laws of causal inference A Gentle Introduction 3. From

369 views • 12 slides
Evidential and Causal Reasoning Much reasoning in AI can be seen as evidential reasoning ,

Evidential and Causal Reasoning Much reasoning in AI can be seen as evidential reasoning ,

Evidential and Causal Reasoning Much reasoning in AI can be seen as evidential reasoning , (observations to a theory) followed by causal reasoning (theory to predictions). Diagnosis Given symptoms, evidential reasoning leads to hypotheses about

243 views • 8 slides
Bayesian Networks 2 Recap of last lecture: Modeling causal relationships with Bayes nets Direct

Bayesian Networks 2 Recap of last lecture: Modeling causal relationships with Bayes nets Direct

Artificial Intelligence 15-381 Mar 29, 2007 Bayesian Networks 2 Recap of last lecture: Modeling causal relationships with Bayes nets Direct cause Indirect cause Common cause Common effect A A A A B B B B C C C P(B|A) P(B|A)

439 views • 22 slides
Causal inference Gary Goertz Kroc Institute for International Peace Studies University of Notre

Causal inference Gary Goertz Kroc Institute for International Peace Studies University of Notre

Causal inference Gary Goertz Kroc Institute for International Peace Studies University of Notre Dame ggoertz@nd.edu Spring 2018 Experiments are the gold standard for causal inference Strategies of causal inference are satisfactory (with

318 views • 11 slides
Outline Inference in Bayes Nets Variable Elimination Bayes Nets (cont) CS 486/686

Outline Inference in Bayes Nets Variable Elimination Bayes Nets (cont) CS 486/686

Outline Inference in Bayes Nets Variable Elimination Bayes Nets (cont) CS 486/686 University of Waterloo May 30, 2006 2 CS486/686 Lecture Slides (c) 2006 C. Boutilier, P. Poupart & K. Larson Simple Forward Inference (Chain)

226 views • 5 slides
Political Science 209 - Fall 2018 Causal Inference Florian Hollenbach 7th September 2018 Causal

Political Science 209 - Fall 2018 Causal Inference Florian Hollenbach 7th September 2018 Causal

Political Science 209 - Fall 2018 Causal Inference Florian Hollenbach 7th September 2018 Causal Inference What do you think is causal inference? Florian Hollenbach 1 Causal Inference causal: relationship between things where one causes

759 views • 47 slides
A Brief Introduction to Causal Inference Brady Neal causalcourse.com What is causal inference?

A Brief Introduction to Causal Inference Brady Neal causalcourse.com What is causal inference?

A Brief Introduction to Causal Inference Brady Neal causalcourse.com What is causal inference? Inferring the effects of any treatment/policy/intervention/etc. 2 / 28 Brady Neal What is causal inference? Inferring the effects of any

2.8k views • 102 slides
Causal Inference Theory and Applications Dr. Matthias Uflacker, Johannes Huegle, Christopher

Causal Inference Theory and Applications Dr. Matthias Uflacker, Johannes Huegle, Christopher

Causal Inference Theory and Applications Dr. Matthias Uflacker, Johannes Huegle, Christopher Schmidt April 24, 2018 Agenda April 24, 2018 Jupyter Notebook Causal Inference in Application Recap Causal Inference in a Nutshell

520 views • 33 slides
Causal Inference By: Miguel A. Hern an and James M. Robins Part I: Causal inference without

Causal Inference By: Miguel A. Hern an and James M. Robins Part I: Causal inference without

Causal Inference By: Miguel A. Hern an and James M. Robins Part I: Causal inference without models Chapter 1: A definition of causal effect 27 th November, 2013 27 th November, 2013 Chapter 1 (Hern an & Robins) Definition of causal

491 views • 13 slides
Geographic Data Science - Lecture IX Causal Inference Dani Arribas-Bel Today Correlation Vs

Geographic Data Science - Lecture IX Causal Inference Dani Arribas-Bel Today Correlation Vs

Geographic Data Science - Lecture IX Causal Inference Dani Arribas-Bel Today Correlation Vs Causation Causal inference Why/when causality matters Hurdles to causal inference & strategies to overcome them Correlation Vs Causation

457 views • 26 slides
Geographic Data Science - Lecture IX Causal Inference Dani Arribas-Bel Today Correlation Vs

Geographic Data Science - Lecture IX Causal Inference Dani Arribas-Bel Today Correlation Vs

Geographic Data Science - Lecture IX Causal Inference Dani Arribas-Bel Today Correlation Vs Causation Causal inference Why/when causality matters Hurdles to causal inference & strategies to overcome them Correlation Vs Causation

741 views • 45 slides
Arguing about potential causal relations Leila Amgoud Henri Prade IRIT, Universit de

Arguing about potential causal relations Leila Amgoud Henri Prade IRIT, Universit de

Arguing about potential causal relations Leila Amgoud Henri Prade IRIT, Universit de Toulouse, CNRS Reasoning about causality i) Deductive causal reasoning - generic causal relations - particular situation predict what is going to take

669 views • 19 slides
Probability recap CS 188: Artificial Intelligence Conditional probability Product rule

Probability recap CS 188: Artificial Intelligence Conditional probability Product rule

Probability recap CS 188: Artificial Intelligence Conditional probability Product rule Bayes Nets Chain rule Representation and Independence X, Y independent iff: Pieter Abbeel UC Berkeley X and Y are

276 views • 8 slides
Causal Dependence Boris Kment 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Causal Dependence Boris Kment 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Counterfactual Reasoning and Causal Dependence Boris Kment 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

551 views • 25 slides
Geometric Processing CS 148: Summer 2016 Introduction of Graphics and Imaging Zahid Hossain

Geometric Processing CS 148: Summer 2016 Introduction of Graphics and Imaging Zahid Hossain

Geometric Processing CS 148: Summer 2016 Introduction of Graphics and Imaging Zahid Hossain https://en.wikipedia.org/wiki/Ray_tracing_%28graphics%29 Final Project Milestone 1: Get to the crux of the project by this milestone 5%

1.19k views • 72 slides
Probabilistic Models CS 4100: Artificial Intelligence Bayes Nets Models describe how (a

Probabilistic Models CS 4100: Artificial Intelligence Bayes Nets Models describe how (a

Probabilistic Models CS 4100: Artificial Intelligence Bayes Nets Models describe how (a portion of) the world works ks Mo Model els ar are e al alway ays simplifi ficat cations May not account for every variable May not

268 views • 8 slides
ECE 4524 Artificial Intelligence and Engineering Applications Lecture 19: Bayesian Networks

ECE 4524 Artificial Intelligence and Engineering Applications Lecture 19: Bayesian Networks

ECE 4524 Artificial Intelligence and Engineering Applications Lecture 19: Bayesian Networks Reading: AIAMA 14.1-14.4 Todays Schedule: Bayesian Networks Causal Reasoning Exact Inference Bayesian Networks (BN) A BN is a

206 views • 9 slides
Reflections and Caustics Kaarel T onisson 1/24 Introduction Photons Photons in

Reflections and Caustics Kaarel T onisson 1/24 Introduction Photons Photons in

Reflections and Caustics Kaarel T onisson 1/24 Introduction Photons Photons in physics Photons in computer graphics Fake reflection techniques Caustics Basics Caustics in computer graphics Caustic textures

635 views • 24 slides
Efficient Caustic Rendering with Lightweight Photon Mapping Pascal Grittmann 1,3 Arsne

Efficient Caustic Rendering with Lightweight Photon Mapping Pascal Grittmann 1,3 Arsne

Efficient Caustic Rendering with Lightweight Photon Mapping Pascal Grittmann 1,3 Arsne Prard-Gayot 1 Philipp Slusallek 1,2 Jaroslav Krivnek 3,4 1 Saarland University 2 DFKI Saarbrcken 3 Charles University, Prague 4 Render Legion The Idea

635 views • 30 slides

More recommend