CS344M Autonomous Multiagent Systems Todd Hester Department of - - PowerPoint PPT Presentation

▶
cs344m autonomous multiagent systems
SMART_READER_LITE
LIVE PREVIEW

CS344M Autonomous Multiagent Systems Todd Hester Department of - - PowerPoint PPT Presentation

Mar 04, 2023 11 likes •563 views

CS344M Autonomous Multiagent Systems Todd Hester Department of Computer Science The University of Texas at Austin Good Afternoon, Colleagues Are there any questions? Todd Hester Logistics Next weeks readings Todd Hester Logistics

slide-1
SLIDE 1

CS344M Autonomous Multiagent Systems

Todd Hester Department of Computer Science The University of Texas at Austin

slide-2
SLIDE 2

Good Afternoon, Colleagues

Are there any questions?

Todd Hester

slide-3
SLIDE 3

Logistics

  • Next week’s readings

Todd Hester

slide-4
SLIDE 4

Logistics

  • Next week’s readings
  • Progress reports due just over 2 weeks

Todd Hester

slide-5
SLIDE 5

Logistics

  • Next week’s readings
  • Progress reports due just over 2 weeks

Todd Hester

slide-6
SLIDE 6

Surveys

  • Readings:

− Mostly good, some mixed reviews

Todd Hester

slide-7
SLIDE 7

Surveys

  • Readings:

− Mostly good, some mixed reviews − Tips of what to get from the reading

Todd Hester

slide-8
SLIDE 8

Surveys

  • Readings:

− Mostly good, some mixed reviews − Tips of what to get from the reading − Some people didn’t like long papers

Todd Hester

slide-9
SLIDE 9

Surveys

  • Readings:

− Mostly good, some mixed reviews − Tips of what to get from the reading − Some people didn’t like long papers

  • Writing

− Most like reading responses

Todd Hester

slide-10
SLIDE 10

Surveys

  • Readings:

− Mostly good, some mixed reviews − Tips of what to get from the reading − Some people didn’t like long papers

  • Writing

− Most like reading responses − Most like openness of response

Todd Hester

slide-11
SLIDE 11

Surveys

  • Readings:

− Mostly good, some mixed reviews − Tips of what to get from the reading − Some people didn’t like long papers

  • Writing

− Most like reading responses − Most like openness of response − “By making us think about the readings, either through specific questions or through free-responses, the writing exercises help us to get more involved with the readings.”

Todd Hester

slide-12
SLIDE 12

− “They help me reaffirm my understanding of the reading, and in that sense they are helpful”

Todd Hester

slide-13
SLIDE 13

− “They help me reaffirm my understanding of the reading, and in that sense they are helpful”

  • Classroom discussions:

− Discussion format good

Todd Hester

slide-14
SLIDE 14

− “They help me reaffirm my understanding of the reading, and in that sense they are helpful”

  • Classroom discussions:

− Discussion format good, improving recently

Todd Hester

slide-15
SLIDE 15

− “They help me reaffirm my understanding of the reading, and in that sense they are helpful”

  • Classroom discussions:

− Discussion format good, improving recently − “The discussions are usually thought-provoking.”

Todd Hester

slide-16
SLIDE 16

− “They help me reaffirm my understanding of the reading, and in that sense they are helpful”

  • Classroom discussions:

− Discussion format good, improving recently − “The discussions are usually thought-provoking.” − Better jumping off point for discussion

Todd Hester

slide-17
SLIDE 17

− “They help me reaffirm my understanding of the reading, and in that sense they are helpful”

  • Classroom discussions:

− Discussion format good, improving recently − “The discussions are usually thought-provoking.” − Better jumping off point for discussion

  • Classroom sessions:

− More activities that involve collaboration and design

Todd Hester

slide-18
SLIDE 18

− “They help me reaffirm my understanding of the reading, and in that sense they are helpful”

  • Classroom discussions:

− Discussion format good, improving recently − “The discussions are usually thought-provoking.” − Better jumping off point for discussion

  • Classroom sessions:

− More activities that involve collaboration and design − More machine learning tutorials (ML/GP/NN/etc)

Todd Hester

slide-19
SLIDE 19

− “They help me reaffirm my understanding of the reading, and in that sense they are helpful”

  • Classroom discussions:

− Discussion format good, improving recently − “The discussions are usually thought-provoking.” − Better jumping off point for discussion

  • Classroom sessions:

− More activities that involve collaboration and design − More machine learning tutorials (ML/GP/NN/etc)

  • Programming assignments:

− Good intro to simulator

Todd Hester

slide-20
SLIDE 20

− “They help me reaffirm my understanding of the reading, and in that sense they are helpful”

  • Classroom discussions:

− Discussion format good, improving recently − “The discussions are usually thought-provoking.” − Better jumping off point for discussion

  • Classroom sessions:

− More activities that involve collaboration and design − More machine learning tutorials (ML/GP/NN/etc)

  • Programming assignments:

− Good intro to simulator, but C/C++ syntax is hard

Todd Hester

slide-21
SLIDE 21

− “They help me reaffirm my understanding of the reading, and in that sense they are helpful”

  • Classroom discussions:

− Discussion format good, improving recently − “The discussions are usually thought-provoking.” − Better jumping off point for discussion

  • Classroom sessions:

− More activities that involve collaboration and design − More machine learning tutorials (ML/GP/NN/etc)

  • Programming assignments:

− Good intro to simulator, but C/C++ syntax is hard

Todd Hester

slide-22
SLIDE 22

− “I think the initial assignments to a good job of exposing us to the properties of both simulators.”

Todd Hester

slide-23
SLIDE 23

− “I think the initial assignments to a good job of exposing us to the properties of both simulators.” − Better documentation, specific detail for different functions

Todd Hester

slide-24
SLIDE 24

− “I think the initial assignments to a good job of exposing us to the properties of both simulators.” − Better documentation, specific detail for different functions

  • Projects:

− Good, generally excited about projects

Todd Hester

slide-25
SLIDE 25

− “I think the initial assignments to a good job of exposing us to the properties of both simulators.” − Better documentation, specific detail for different functions

  • Projects:

− Good, generally excited about projects − “Glad we got to formulate our own projects and that direction is given when needed.”

Todd Hester

slide-26
SLIDE 26

− “I think the initial assignments to a good job of exposing us to the properties of both simulators.” − Better documentation, specific detail for different functions

  • Projects:

− Good, generally excited about projects − “Glad we got to formulate our own projects and that direction is given when needed.” − “I look forward to diving into the final project.”

Todd Hester

slide-27
SLIDE 27

− “I think the initial assignments to a good job of exposing us to the properties of both simulators.” − Better documentation, specific detail for different functions

  • Projects:

− Good, generally excited about projects − “Glad we got to formulate our own projects and that direction is given when needed.” − “I look forward to diving into the final project.”

Todd Hester

slide-28
SLIDE 28

My takeaways

− Readings - Give tips on what to focus on

Todd Hester

slide-29
SLIDE 29

My takeaways

− Readings - Give tips on what to focus on − Discussion - Give better initiating questions

Todd Hester

slide-30
SLIDE 30

My takeaways

− Readings - Give tips on what to focus on − Discussion - Give better initiating questions − Activities - More activities getting you designing solutions

Todd Hester

slide-31
SLIDE 31

My takeaways

− Readings - Give tips on what to focus on − Discussion - Give better initiating questions − Activities - More activities getting you designing solutions − Machine learning - More tutorials on machine learning

Todd Hester

slide-32
SLIDE 32

My takeaways

− Readings - Give tips on what to focus on − Discussion - Give better initiating questions − Activities - More activities getting you designing solutions − Machine learning - More tutorials on machine learning − C/C++ help - More advice and specific documentation

Todd Hester

slide-33
SLIDE 33

Game Theory

  • Multiagent systems
  • Economics
  • Social science, law, etc.

Todd Hester

slide-34
SLIDE 34

Goals for Today

  • Understand premises of game theory
  • Understand the notion of utility
  • Understand solution concepts

Todd Hester

slide-35
SLIDE 35

Goals for Today

  • Understand premises of game theory
  • Understand the notion of utility
  • Understand solution concepts

– Dominant strategy – Nash equilibrium – Pareto optimality – Maximum social welfare – Maximin strategy

Todd Hester

slide-36
SLIDE 36

Prisoner’s Dilemma

❈♦❧✉♠♥ ❈✭✶✮ ❉✭✷✮ ❈✭✶✮ ✸✱✸ ✵✱✺ ❘♦✇ ❉✭✷✮ ✺✱✵ ✶✱✶

Todd Hester

slide-37
SLIDE 37

Game Theory Premises

  • Simultaneous actions
  • No communication
  • Outcome depends on combination of actions

Todd Hester

slide-38
SLIDE 38

Game Theory Premises

  • Simultaneous actions
  • No communication
  • Outcome depends on combination of actions
  • Utility (payoff) encapsulates everything about preferences
  • ver outcomes

Todd Hester

slide-39
SLIDE 39

Utility

  • Money is a useful analogy for utility

− But they’re not equivalent

Todd Hester

slide-40
SLIDE 40

Utility

  • Money is a useful analogy for utility

− But they’re not equivalent

  • Diminishing values

Todd Hester

slide-41
SLIDE 41

Utility

  • Money is a useful analogy for utility

− But they’re not equivalent

  • Diminishing values
  • Risk aversion

Todd Hester

slide-42
SLIDE 42

Utility

  • Money is a useful analogy for utility

− But they’re not equivalent

  • Diminishing values
  • Risk aversion
  • Loss aversion

Todd Hester

slide-43
SLIDE 43

Utility

  • Money is a useful analogy for utility

− But they’re not equivalent

  • Diminishing values
  • Risk aversion
  • Loss aversion
  • Friendliness/vindictiveness

Todd Hester

slide-44
SLIDE 44

Solution Concepts

  • Dominant strategy

Todd Hester

slide-45
SLIDE 45

Solution Concepts

  • Dominant strategy
  • Nash equilibrium

Todd Hester

slide-46
SLIDE 46

Solution Concepts

  • Dominant strategy
  • Nash equilibrium
  • Pareto optimality

Todd Hester

slide-47
SLIDE 47

Solution Concepts

  • Dominant strategy
  • Nash equilibrium
  • Pareto optimality
  • Maximum social welfare

Todd Hester

slide-48
SLIDE 48

Solution Concepts

  • Dominant strategy
  • Nash equilibrium
  • Pareto optimality
  • Maximum social welfare
  • Maximin strategy

Todd Hester

slide-49
SLIDE 49

Prisoner’s Dilemma

❈♦❧✉♠♥ ❈✭✶✮ ❉✭✷✮ ❈✭✶✮ ✸✱✸ ✵✱✺ ❘♦✇ ❉✭✷✮ ✺✱✵ ✶✱✶

Todd Hester

slide-50
SLIDE 50

Chicken

❈♦❧✉♠♥ ❈✭✶✮ ❉✭✷✮ ❈✭✶✮ ✸✱✸ ✶✱✺ ❘♦✇ ❉✭✷✮ ✺✱✶ ✵✱✵

Todd Hester

slide-51
SLIDE 51

Discussion

  • How would you design an agent to play iterated prisoner’s

dilemma?

Todd Hester

slide-52
SLIDE 52

Discussion

  • How would you design an agent to play iterated prisoner’s

dilemma?

  • What if you could submit up to 10 entries to the

tournament?

Todd Hester

slide-53
SLIDE 53

Discussion

  • How would you design an agent to play iterated prisoner’s

dilemma?

  • What if you could submit up to 10 entries to the

tournament?

  • What is your strategy if you are playing only one game

with a stranger, but can communicate first?

Todd Hester

slide-54
SLIDE 54

Discussion

  • How would you design an agent to play iterated prisoner’s

dilemma?

  • What if you could submit up to 10 entries to the

tournament?

  • What is your strategy if you are playing only one game

with a stranger, but can communicate first?

Todd Hester

slide-55
SLIDE 55

Split or Steal

❈♦❧✉♠♥ ❙♣❧✐t✭✶✮ ❙t❡❛❧✭✷✮ ❙♣❧✐t✭✶✮ ✵✳✺✱✵✳✺ ✵✱✶ ❘♦✇ ❙t❡❛❧✭✷✮ ✶✱✵ ✵✱✵

Todd Hester