Making Intelligence

Loading...

Flash Player 9 (or above) is needed to view presentations.
We have detected that you do not have it on your computer. To install it, go here.

0 comments

Post a comment

    Post a comment
    Embed Video
    Edit your comment Cancel

    1 Event

    Making Intelligence - Presentation Transcript

    1. Artificial Intelligence Joseph Gentle
    2. Ability to make optimising choices Intelligence
    3. Input Output Intelligence Reward
    4. Input Output Intelligence Reward
    5. Input Input Input Input Output Output Output Intelligence Output Intelligence Intelligence Intelligence Reward Reward Reward Reward
    6. Input Output Intelligence Reward
    7. Observations Actions Intelligence Reward
    8. Observations Actions Intelligence Goal
    9. Input Output Intelligence Reward
    10. Input Output Choice Reward
    11. Input Guided Output Choice Reward
    12. Ability to make optimising choices
    13. Ability to make optimising choices een etw eb oos uts Ch utp o
    14. Ability to make optimising choices een etw eb oos uts Ch WRT Reward Function utp o
    15. Input Guided Output Choice R T Maximise R(t) t=0
    16. Input Guided Output Choice R T Maximise R(t) dt 0
    17. Input Guided Output Choice R T Maximise t R(t)γ dt 0 γ ≈ 0.99
    18. Examples
    19. Chess Player Chess board Move Chess Player Victory?
    20. Vacuum Cleaner Sensors Vacuum Motors Cleaner AI Cleanliness of floor - run out of power?
    21. Clean my Write a Play Chess floor sonnet
    22. Clean my Write a Play Chess floor sonnet Chess Vacuum Sonnet Player Cleaner AI writer
    23. Clean my Write a Play Chess floor sonnet Chess Vacuum Sonnet Player Cleaner AI writer Generalisable Intelligence
    24. Philosophy
    25. Input Output Intelligence Reward
    26. Output Intelligence Reward
    27. Input Output Intelligence
    28. Input Random! Intelligence
    29. Input Intelligence Reward
    30. Input Choice?? Reward
    31. Input Intelligence
    32. Intelligence World Input
    33. Understanding Intelligence World Input
    34. Prediction Intelligence World Input
    35. Input Intelligence Output Reward
    36. Input Intelligence Output ∞
    37. Input Intelligence Random! ∞
    38. People
    39. Input Output Intelligence Reward
    40. Ability to make optimising choices Input Output Intelligence Reward
    41. Input Output Choice Reward
    42. Brain Output Input Choice Reward
    43. Brain Muscle control Senses Choice Reward
    44. Brain Muscle control Senses Choice Instantaneous long term sum of instantanious reward Happiness
    45. Brain Muscle control Senses Choice Reward
    46. Brain Muscle control Senses Choice Reward
    47. Brain Conscious Muscle control Senses Choice Reward Subconscious
    48. Brain Conscious Muscle control Senses Choice Reward Subconscious
    49. Brain Conscious Muscle control Senses Choice ∞ Subconscious
    50. Ability to make optimising choices Input Output Intelligence Reward
    51. Bias Its a good thing.
    52. 1, 2, 3, ? 2, 4, 8, ? 2, 4, 6, ?
    53. 1, 2, 3, ?
    54. 4 1, 2, 3, ? Increasing numbers
    55. 5 Fibonacci sequence 4 1, 2, 3, ? Increasing numbers
    56. 5 Fibonacci sequence 4 1, 2, 3, ? Increasing numbers 2 Largest prime dividing N
    57. 5 5 Square-free Fibonacci numbers sequence 4 6 1, 2, 3, ? Increasing Central binomial numbers coefficients 10 2 Numbers with no Largest prime carry when squared dividing N
    58. Source: Online Encyclopedia of Integer Sequences
    59. 5 5 Square-free Fibonacci numbers sequence 4 6 1, 2, 3, ? Increasing Central binomial numbers coefficients 10 2 Numbers with no Largest prime carry when squared dividing N
    60. 5 5 Square-free Fibonacci numbers sequence 4 1, 2, 3, ? Increasing numbers 6 10 2 Central binomial Numbers with no Largest prime coefficients carry when squared dividing N
    61. 4 1, 2, 3, ? Increasing numbers
    62. 1, 2, 3, 4 is the simplest pattern 4 1, 2, 3, ? Increasing numbers
    63. 1, 2, 3, 4 is the simplest pattern 4 1, 2, 3, ? Increasing numbers 1, 2, 3, 4 is a common sequence
    64. 1, 2, 3, 4 is the Simplicity simplest pattern 1, 2, 3, 4 is a Induction common sequence
    65. 1, 2, 3, 4 is the Simplicity simplest pattern The simplest explanation is usually the best. Ockham’s Razor
    66. Patterns repeat Induction 1, 2, 3, 4 is a Induction common sequence
    67. Things we’ve seen New things before Simplicity Induction
    68. Simplicity 1, 2, 3, ? 5 5 Square-free Fibonacci > 4 numbers sequence 6 Increasing numbers Central binomial coefficients 10 2 Numbers with no Largest prime carry when squared dividing N
    69. Simplicity 1, 2, 3, ? Largest prime dividing the number’s position in 2 the sequence 5 Numbers in the Fibonacci sequence from n = 2 5 Natural numbers which aren’t perfect squares 4 Natural numbers 6 Central binomial coefficients of C(n, f loor(n/2)) 10 Natural numbers with no carry when squared
    70. Simplicity 1, 2, 3, ? 4 Natural numbers 10 Natural numbers with no carry when squared 5 Numbers in the Fibonacci sequence from n = 2 5 Natural numbers which aren’t perfect squares 6 Central binomial coefficients of C(n, f loor(n/2)) Largest prime dividing the number’s position in 2 the sequence
    71. Simplicity Length 15 Natural numbers 43 Natural numbers with no carry when squared 44 Numbers in the Fibonacci sequence from n = 2 46 Natural numbers which aren’t perfect squares 49 Central binomial coefficients of C(n, f loor(n/2)) Largest prime dividing the number’s position in 62 the sequence
    72. The length of the shortest generating program in a given language Kolmogorov Complexity
    73. Simplicity Central binomial coefficients of C(n, f loor(n/2)) 49 Largest prime dividing the number’s position in 62 the sequence
    74. Simplicity Central binomial coefficients of C(n, f loor(n/2)) 49 Largest prime dividing the number’s position in 62 the sequence
    75. Simplicity 15 Natural numbers Natural numbers with no carry when squared 43 Numbers in the Fibonacci sequence from n = 2 44 Natural numbers which aren’t perfect squares 46 Central binomial coefficients of C(n, f loor(n/2)) 49 Largest prime dividing the number’s position in 62 the sequence
    76. Architecture
    77. Brain Senses Motor Control
    78. Brain Senses Motor Control
    79. Brain Mapping Senses Vision Write Sonnets Hearing Play chess Motor Taste Control Choice Reward function
    80. Born with Learned Vision Write Choice Sonnets Hearing Taste Reward Play chess function Mapping
    81. Born with Learned Vision Write Choice Sonnets Hearing Taste Reward Play chess function Mapping
    82. Senses Motor Control Brain These are NOT the same Take inspiration, not knowledge! Input Output Intelligence Reward
    83. Intelligence Input Output Reward .... and remember you can design this however you want!
    84. Neural Net Intelligence Input Output Reward
    85. Robotics Intelligence Input Vision Output Mapping Choice function Value function Reward
    86. Robotics approach is hard because you have to engineer all the behaviours Born with Learned Nerural nets approach is hard because your neural net needs to be able of learning everything. Robotics Neural Net Vision Mapping Choice Value function function
    87. Born with Learned Robotics Neural Net Vision Mapping Choice Value function function
    88. Clean my Write a Play Chess floor sonnet Chess Vacuum Sonnet Player Cleaner AI writer Generalisable Intelligence
    89. My Architecture
    90. Intelligence Input Input model World model
    91. Intelligence Input Input model World model Value function Reward
    92. Intelligence Input Input Action model model World model Value function Reward
    93. Intelligence Input Input Action model model World model Value function Output Reward Choice function
    94. Intelligence Input Input model World model
    95. Input Intelligence
    96. Intelligence World Input
    97. Understanding Intelligence World Input
    98. Prediction Intelligence World Input
    99. Prediction Intelligence World model World Input Input model
    100. Prediction Intelligence World model World Input Input model Prediction Accuracy
    101. Input Intelligence Predictions World model Input model Prediction Accuracy
    102. My Bias
    103. Prediction Intelligence World model World Input Input model
    104. World model = World Input model
    105. Prediction Intelligence World model UNKNOWN World Input Input model
    106. World Intelligence World model World Input Input model World
    107. Intelligence World model 1, 2, 3, ? Input
    108. 5 Fibonacci sequence Intelligence 4 World model 1, 2, 3, ? Increasing Input numbers 2 Largest prime dividing N
    109. Intelligence World model 1, 2, 3, 4 Input Increasing numbers
    110. Simplicity Intelligence World model 1, 2, 3, 4 Input Increasing numbers Induction
    111. 1, 2, 3, 4 is the Simplicity simplest pattern 1, 2, 3, 4 is a Induction common sequence
    112. Simplicity Induction
    113. Small model > Simplicity Big model Induction
    114. Small model > Simplicity Big model Induction Learn patterns
    115. Input Intelligence Predictions World model Input model Prediction Accuracy
    116. Input Intelligence Predictions World model Input model Prediction Accuracy Learn patterns Prefer small models
    117. @sineltor http://seph.pbwiki.com/

    + sineltorsineltor, 7 months ago

    custom

    438 views, 0 favs, 0 embeds more stats

    Making intelligence. Talk given at Barcamp Canberra more

    More info about this document

    © All Rights Reserved

    Go to text version

    • Total Views 438
      • 438 on SlideShare
      • 0 from embeds
    • Comments 0
    • Favorites 0
    • Downloads 3
    Most viewed embeds

    more

    All embeds

    less

    Flagged as inappropriate Flag as inappropriate
    Flag as inappropriate

    Select your reason for flagging this presentation as inappropriate. If needed, use the feedback form to let us know more details.

    Cancel
    File a copyright complaint
    Having problems? Go to our helpdesk?

    Categories

    Groups / Events