probabilistic programming in birch
play

Probabilistic Programming in Birch www.birch-lang.org Lawrence - PowerPoint PPT Presentation

May 31, 2023 •387 likes •1.88k views

Probabilistic Programming in Birch www.birch-lang.org Lawrence Murray Department of Information Technology, Uppsala University Outline 1. Graphical models probabilistic programs. 2. Birch: motivation and design. 3. Birch: language

  1. Methods in Birch Inference methods are also written in the Birch language. ▶ Currently available are: ▶ Analytical solutions ▶ Importance sampling ▶ Bootstrap particle filter ▶ Alive particle filter ▶ Auxiliary particle filter (automated) ▶ Rao–Blackwellized particle filter (automated) ▶ Not far off are: ▶ Particle MCMC methods ▶ Other MCMC methods. Lawrence Murray 19 / 30

  2. 3 Birch: language features Lawrence Murray 20 / 30

  3. Optionals Optionals allow variables to have a value of a particular type, or no value at all. ▶ They are used in other programming languages (e.g. Swift) to eliminate boilerplate that checks for null values, e.g. a function checking its arguments. ▶ In Birch, they are used for the same purpose, but also a second role: to represent missing values . Lawrence Murray 21 / 30

  4. Randoms Randoms are optionals to which a probability distribution can be attached. ▶ When they don’t have a value , the probability distribution can be used to automatically simulate a value . ▶ Once a random has a value, that value is final, it cannot be overwritten. Lawrence Murray 22 / 30

  5. Delayed sampling ▶ Randoms are essential for the delayed sampling mechanism within Birch. ▶ This is a heuristic algorithm for performing analytical optimizations at runtime. ▶ It automatically yields optimizations such as variable elimination/collapsing, Rao–Blackwellization and locally-optimal proposals. See: L. M. Murray, D. Lundén, J. Kudlicka, D. Broman, and T. B. Schön. Delayed sam- pling and automatic Rao–Blackwellization of probabilistic programs. Proceedings of the 21st International Conference on Artificial Intelligence and Statistics (AISTATS), 2018. URL https://arxiv.org/abs/1708.07787 Lawrence Murray 23 / 30

  6. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { y[n] ~ Gaussian(x, 1.0); } stdout.print(x); Lawrence Murray 24 / 30

  7. Delayed sampling example Code Checkpoint assume x x ~ Gaussian(0.0, 1.0); for (n in 1..N) { y[n] ~ Gaussian(x, 1.0); } stdout.print(x); x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  8. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { y[n] ~ Gaussian(x, 1.0); } stdout.print(x); x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  9. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  10. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  11. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  12. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 1 x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  13. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 1 x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  14. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 1 x 1 y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  15. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 2 x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  16. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 2 x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  17. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 2 x 2 y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  18. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 3 x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  19. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 3 x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  20. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 3 x 3 y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  21. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 4 x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  22. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 4 x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  23. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 4 x 4 y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  24. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { observe y[n] y[n] ~ Gaussian(x, 1.0); } stdout.print(x); 5 x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  25. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { y[n] ~ Gaussian(x, 1.0); } value x stdout.print(x); 5 x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  26. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { y[n] ~ Gaussian(x, 1.0); } stdout.print(x); x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  27. Delayed sampling example Code Checkpoint x ~ Gaussian(0.0, 1.0); for (n in 1..N) { y[n] ~ Gaussian(x, 1.0); } stdout.print(x); x y[1] y[2] y[3] y[4] y[5] Lawrence Murray 24 / 30

  28. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); Lawrence Murray 25 / 30

  29. Delayed sampling Code Checkpoint assume x[1] x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  30. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); observe y[1] y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  31. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); observe y[1] y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  32. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); observe y[1] y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  33. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); observe y[1] y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  34. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { assume x[t] x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  35. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  36. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 1 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  37. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 1 x[1] x[2] x[3] x[4] x[5] 1 y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  38. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  39. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { assume x[t] x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  40. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  41. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 2 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  42. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 2 x[1] x[2] x[3] x[4] x[5] 2 y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  43. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 3 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  44. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { assume x[t] x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 3 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  45. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 3 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  46. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 3 3 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  47. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 3 3 x[1] x[2] x[3] x[4] x[5] 3 y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  48. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 3 4 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  49. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { assume x[t] x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 3 4 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  50. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 3 4 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  51. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 3 4 4 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  52. Delayed sampling Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 3 4 4 x[1] x[2] x[3] x[4] x[5] 4 y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  53. Delayed sampling: Kalman Filter Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); observe y[t] y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); 1 2 3 4 5 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  54. Delayed sampling: Kalman Filter Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } value x[1] stdout.print(x[1]); 1 2 3 5 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  55. Delayed sampling: Kalman Filter Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } value x[1] stdout.print(x[1]); 1 2 5 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  56. Delayed sampling: Kalman Filter Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } value x[1] stdout.print(x[1]); 1 5 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  57. Delayed sampling: Kalman Filter Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } value x[1] stdout.print(x[1]); 5 x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  58. Delayed sampling: Kalman Filter Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } value x[1] stdout.print(x[1]); x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  59. Delayed sampling: Kalman Filter Code Checkpoint x[1] ~ Gaussian(0.0, 1.0); y[1] ~ Gaussian(x[1], 1.0); for (t in 2..T) { x[t] ~ Gaussian(a*x[t - 1], 1.0); y[t] ~ Gaussian(x[t], 1.0); } stdout.print(x[1]); x[1] x[2] x[3] x[4] x[5] y[1] y[2] y[3] y[4] y[5] Lawrence Murray 25 / 30

  60. Delayed sampling Lawrence Murray 26 / 30

  61. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

  62. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

  63. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

  64. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

  65. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

  66. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

  67. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

  68. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

  69. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

  70. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

  71. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] 1 x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

  72. Delayed sampling x_n[1] x_n[2] x_n[3] x_n[4] x_n[5] 1 x_l[1] x_l[2] x_l[3] x_l[4] x_l[5] y_l[1] y_l[2] y_l[3] y_l[4] y_l[5] y_n[1] y_n[2] y_n[3] y_n[4] y_n[5] Lawrence Murray 26 / 30

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

Automated learning with a probabilistic programming language: Birch 1. The Birch probabilistic

Automated learning with a probabilistic programming language: Birch 1. The Birch probabilistic

Lawrence Murray Department of Information Technology, Uppsala University Outline Automated learning with a probabilistic programming language: Birch 1. The Birch probabilistic programming language. 2. The delayed sampling heuristic. 3. A

1.43k views • 142 slides
CS 4110 Probabilistic Programming Probabilistic Programming It's not about writing software.

CS 4110 Probabilistic Programming Probabilistic Programming It's not about writing software.

CS 4110 Probabilistic Programming Probabilistic Programming It's not about writing software. Probabilistic Programming Probabilistic programming is a tool for statistical modeling. OR A probabilistic programming language is a plain old

688 views • 41 slides
Reactive Probabilistic Programming Semantics with Mixed Nondeterministic/Probabilistic Automata

Reactive Probabilistic Programming Semantics with Mixed Nondeterministic/Probabilistic Automata

Reactive Probabilistic Programming Semantics with Mixed Nondeterministic/Probabilistic Automata Albert Benveniste Jean-Baptiste Raclet INRIA Rennes and IRIT Toulouse September 18, 2020 What is Probabilistic Programming? Bringing the

855 views • 59 slides
A Brief Introduction to Probabilistic and Quantum Programming Part II Ugo Dal Lago Universidade

A Brief Introduction to Probabilistic and Quantum Programming Part II Ugo Dal Lago Universidade

A Brief Introduction to Probabilistic and Quantum Programming Part II Ugo Dal Lago Universidade do Minho, Friday May 5, 2017 Section 1 Probabilistic Programming Languages Flipping a Coin Making any programming language probabilistic is

1.03k views • 92 slides
Discrete Probabilistic Programming from First Principles Guy Van den Broeck The 6 th Workshop on

Discrete Probabilistic Programming from First Principles Guy Van den Broeck The 6 th Workshop on

Discrete Probabilistic Programming from First Principles Guy Van den Broeck The 6 th Workshop on Probabilistic Logic Programming (PLP) Sep 21, 2019 What are probabilistic programs? What is the formal semantics? How to do exact inference? What

830 views • 61 slides
Introduction to Probabilistic and Quantum Programming Part II Ugo Dal Lago BISS 2014, Bertinoro

Introduction to Probabilistic and Quantum Programming Part II Ugo Dal Lago BISS 2014, Bertinoro

Introduction to Probabilistic and Quantum Programming Part II Ugo Dal Lago BISS 2014, Bertinoro Section 1 Probabilistic Programming Languages Flipping a Coin Adding probabilistic behaviour to any programming language is relatively easy ,

459 views • 24 slides
BIRCH STREET PLAZA: ROSLINDALE VILLAGE Birch St Project Team Birch St Process (18 months)

BIRCH STREET PLAZA: ROSLINDALE VILLAGE Birch St Project Team Birch St Process (18 months)

BIRCH STREET PLAZA: ROSLINDALE VILLAGE Birch St Project Team Birch St Process (18 months) Design City Site Selection RFQ Process Approvals Small Task Force Contacted City staff and City of Boston met ABC put out RFQ to

136 views • 12 slides
Discrete Probabilistic Programming from First Principles Guy Van den Broeck The Fourth

Discrete Probabilistic Programming from First Principles Guy Van den Broeck The Fourth

Discrete Probabilistic Programming from First Principles Guy Van den Broeck The Fourth International Workshop on Declarative Learning Based Programming (DeLBP) Aug 11, 2019 What are probabilistic programs? What is the formal semantics? How

1.06k views • 63 slides
Edward: Deep Probabilistic Programming Extended Seminar Systems and Machine Learning Steven

Edward: Deep Probabilistic Programming Extended Seminar Systems and Machine Learning Steven

Edward: Deep Probabilistic Programming Extended Seminar Systems and Machine Learning Steven Lang 13.02.2020 1 Outline Introduction Refresher on Probabilistic Modeling Deep Probabilistic Programming Compositional Representations in

912 views • 58 slides
Theres Something About Bayes Effective Probabilistic Programming for the Rest of Us James

Theres Something About Bayes Effective Probabilistic Programming for the Rest of Us James

Theres Something About Bayes Effective Probabilistic Programming for the Rest of Us James Bornholt Todd Mytkowicz Microsoft Research Kathryn S. McKinley Programs are doing (probabilistic) inference, even if they dont realise it.

548 views • 13 slides
Approximated Probabilistic Answer Set Programming Eduardo Menezes de Morais Department of

Approximated Probabilistic Answer Set Programming Eduardo Menezes de Morais Department of

Motivation ASP PSAT Probabilistic Answer Set Programming Resolution Methods Approximations Future work Approximated Probabilistic Answer Set Programming Eduardo Menezes de Morais Department of Computer Science Universidade de S ao

808 views • 48 slides
The Joy of Probabilistic Answer Set Programming Fabio G. Cozman Universidade de S ao Paulo

The Joy of Probabilistic Answer Set Programming Fabio G. Cozman Universidade de S ao Paulo

The Joy of Probabilistic Answer Set Programming Fabio G. Cozman Universidade de S ao Paulo Goal: to show that the credal semantics for Probabilistic Answer Set Programming (PASP) leads to a very useful modeling language. Answer set

375 views • 14 slides
Probabilistic Programming Fun but Intricate Too! Joost-Pieter Katoen with Friedrich Gretz, Nils

Probabilistic Programming Fun but Intricate Too! Joost-Pieter Katoen with Friedrich Gretz, Nils

Probabilistic Programming is Fun, but Intricate Too Probabilistic Programming Fun but Intricate Too! Joost-Pieter Katoen with Friedrich Gretz, Nils Jansen, Benjamin Kaminski Christoph Matheja, Federico Olmedo and Annabelle McIver Mysore

775 views • 77 slides
Introduction to probabilistic programming Frank Wood fwood@cs.ubc.ca Objectives For Today Get

Introduction to probabilistic programming Frank Wood fwood@cs.ubc.ca Objectives For Today Get

Introduction to probabilistic programming Frank Wood fwood@cs.ubc.ca Objectives For Today Get you to Understand what probabilistic programming is Think generatively Understand inference Importance sampling SMC MCMC

1.25k views • 82 slides
Principles of Probabilistic Programming Lectures at EWSCS 2020 Winter School Joost-Pieter Katoen

Principles of Probabilistic Programming Lectures at EWSCS 2020 Winter School Joost-Pieter Katoen

Principles of Probabilistic Programming Lectures at EWSCS 2020 Winter School Joost-Pieter Katoen EWSCS 2020, Palmse, Estonia Joost-Pieter Katoen Principles of Probabilistic Programming 1/222 What is probabilistic programming? Probabilistic

885 views • 56 slides
Paradoxes of probabilistic programming and how to condition on events of measure zero with

Paradoxes of probabilistic programming and how to condition on events of measure zero with

Paradoxes of probabilistic programming and how to condition on events of measure zero with infinitesimal probabilities (to appear at POPL21) Jules Jacobs Radboud University Nijmegen julesjacobs@gmail.com November 23, 2020 1 / 25

898 views • 75 slides
SUMMARY OF RECOMMENDATIONS AND SDG TARGETS FOR FUTURE REPORTING Speaker: Dr. Gillian Sparkes,

SUMMARY OF RECOMMENDATIONS AND SDG TARGETS FOR FUTURE REPORTING Speaker: Dr. Gillian Sparkes,

SUMMARY OF RECOMMENDATIONS AND SDG TARGETS FOR FUTURE REPORTING Speaker: Dr. Gillian Sparkes, Commissioner for Environmental Sustainability August, 2019 Cultural Landscape Health and Management 1. That the Victorian Government, in

410 views • 22 slides
Developing Dorset Councils Plan September 2019 Dorset Council was formed from 6 previous

Developing Dorset Councils Plan September 2019 Dorset Council was formed from 6 previous

Developing Dorset Councils Plan September 2019 Dorset Council was formed from 6 previous county, district and borough councils on 1 April 2019. Why did we come together to form a new unitary Council? To protect frontline services by

313 views • 13 slides
securing the value of nature Overview A healthy, properly functioning natural environment is

securing the value of nature Overview A healthy, properly functioning natural environment is

NATURAL ENVIRONMENT WHITE PAPER The Natural Choice: securing the value of nature Overview A healthy, properly functioning natural environment is the foundation of sustained economic growth, prospering communities and personal wellbeing.

265 views • 9 slides
From the Birch and Swinnerton Dyer Conjecture to the GL 2 Main Conjecture for elliptic curves by

From the Birch and Swinnerton Dyer Conjecture to the GL 2 Main Conjecture for elliptic curves by

X International Workshop on Differential Equations, Number Theory, Data Analysis Methods and Geometry University of Havana, February 19-23, 2007 From the Birch and Swinnerton Dyer Conjecture to the GL 2 Main Conjecture for elliptic curves by

549 views • 32 slides
Support Vector Machines Alex Leblang and Sam Birch ML Framework Data projected into feature

Support Vector Machines Alex Leblang and Sam Birch ML Framework Data projected into feature

Support Vector Machines Alex Leblang and Sam Birch ML Framework Data projected into feature space, each feature is a dimension. Via Wikipedia, scikit-learn examples SVMs Supervised learning model: train / test Binary, discriminative

350 views • 12 slides
Number Theory and Representation Theory A conference in honor of the 60th birthday of Benedict

Number Theory and Representation Theory A conference in honor of the 60th birthday of Benedict

Number Theory and Representation Theory A conference in honor of the 60th birthday of Benedict Gross Harvard University, Cambridge June 2010 Elliptic curves over real quadratic fields, and the Birch and Swinnerton-Dyer conjecture ... A

1.59k views • 103 slides
Hierarchy An arrangement or classification of things according to inclusiveness A natural

Hierarchy An arrangement or classification of things according to inclusiveness A natural

Hierarchy An arrangement or classification of things according to inclusiveness A natural way of abstraction, summarization, compression, and simplification for understanding Typical setting: organize a given set of objects to a

428 views • 24 slides
CURE: An Efficient Clustering Algorithm for Large Databases Sudipto Guha Rajeev Rastogi Kyuseok

CURE: An Efficient Clustering Algorithm for Large Databases Sudipto Guha Rajeev Rastogi Kyuseok

20171120 CURE: An Efficient Clustering Algorithm for Large Databases Sudipto Guha Rajeev Rastogi Kyuseok Shim Stanford University Bell Laboratories Bell Laboratories Presented by Afnan Ahmad Monday, November 20, 2017 1 Overview of

297 views • 12 slides

More recommend