CS344M Autonomous Multiagent Systems Patrick MacAlpine Department - - PowerPoint PPT Presentation

▶
cs344m autonomous multiagent systems
SMART_READER_LITE
LIVE PREVIEW

CS344M Autonomous Multiagent Systems Patrick MacAlpine Department - - PowerPoint PPT Presentation

Feb 01, 2023 447 likes •1.29k views

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

slide-1
SLIDE 1

CS344M Autonomous Multiagent Systems

Patrick MacAlpine Department of Computer Science The University of Texas at Austin

slide-2
SLIDE 2

Good Afternoon, Colleagues

Are there any questions?

Patrick MacAlpine

slide-3
SLIDE 3

Logistics

  • Next week’s readings in flux

Patrick MacAlpine

slide-4
SLIDE 4

Logistics

  • Next week’s readings in flux
  • Final projects due in 2 weeks!

Patrick MacAlpine

slide-5
SLIDE 5

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

Patrick MacAlpine

slide-6
SLIDE 6

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • What’s the value of the flash?

Patrick MacAlpine

slide-7
SLIDE 7

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • What’s the value of the flash?

− Auctions are simultaneous − Auctions are independent (no combinatorial bids)

Patrick MacAlpine

slide-8
SLIDE 8

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • What’s the value of the flash?

− Auctions are simultaneous − Auctions are independent (no combinatorial bids)

  • ∈ [10, 50] — Depends on the price of the camera

Patrick MacAlpine

slide-9
SLIDE 9

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

Patrick MacAlpine

slide-10
SLIDE 10

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $80

Patrick MacAlpine

slide-11
SLIDE 11

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $80

− score(G∗ f) =

Patrick MacAlpine

slide-12
SLIDE 12

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $80

− score(G∗ f) = max{100 − 80, 10 − 0} = 20

Patrick MacAlpine

slide-13
SLIDE 13

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $80

− score(G∗ f) = max{100 − 80, 10 − 0} = 20 − score(G∗ no-f) =

Patrick MacAlpine

slide-14
SLIDE 14

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $80

− score(G∗ f) = max{100 − 80, 10 − 0} = 20 − score(G∗ no-f) = max{50 − 80, 0 − 0} = 0

Patrick MacAlpine

slide-15
SLIDE 15

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $80

− score(G∗ f) = max{100 − 80, 10 − 0} = 20 − score(G∗ no-f) = max{50 − 80, 0 − 0} = 0 − So value(flash) = 20 − 0 = $20

Patrick MacAlpine

slide-16
SLIDE 16

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $80

− score(G∗ f) = max{100 − 80, 10 − 0} = 20 − score(G∗ no-f) = max{50 − 80, 0 − 0} = 0 − So value(flash) = 20 − 0 = $20

  • Already bought camera ⇒ price = $0

Patrick MacAlpine

slide-17
SLIDE 17

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $80

− score(G∗ f) = max{100 − 80, 10 − 0} = 20 − score(G∗ no-f) = max{50 − 80, 0 − 0} = 0 − So value(flash) = 20 − 0 = $20

  • Already bought camera ⇒ price = $0⇒

value(flash) = 100 − 50 = $50

Patrick MacAlpine

slide-18
SLIDE 18

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

Patrick MacAlpine

slide-19
SLIDE 19

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $20, flash = $10

− value(flash) would be

Patrick MacAlpine

slide-20
SLIDE 20

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $20, flash = $10

− value(flash) would be 80 − 30 = $50 − value(camera) would be

Patrick MacAlpine

slide-21
SLIDE 21

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $20, flash = $10

− value(flash) would be 80 − 30 = $50 − value(camera) would be 90 − 0 = $90

  • But what if prices jump at the end?

Patrick MacAlpine

slide-22
SLIDE 22

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $20, flash = $10

− value(flash) would be 80 − 30 = $50 − value(camera) would be 90 − 0 = $90

  • But what if prices jump at the end?

− Let average past camera price = $80, flash = $30

Patrick MacAlpine

slide-23
SLIDE 23

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • Let current camera price = $20, flash = $10

− value(flash) would be 80 − 30 = $50 − value(camera) would be 90 − 0 = $90

  • But what if prices jump at the end?

− Let average past camera price = $80, flash = $30 − value(flash) = $20 − value(camera) = $70

Patrick MacAlpine

slide-24
SLIDE 24

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

Patrick MacAlpine

slide-25
SLIDE 25

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • What’s the value of the flash?

− Camera price = $70 ⇒ value(flash) = $30 − Camera price = $20 ⇒ value(flash) = $50 − Camera price = $40 ⇒ value(flash) = $50

Patrick MacAlpine

slide-26
SLIDE 26

Bidding for Multiple Items

utility camera alone $50 flash alone 10 both 100 neither

  • What’s the value of the flash?

− Camera price = $70 ⇒ value(flash) = $30 − Camera price = $20 ⇒ value(flash) = $50 − Camera price = $40 ⇒ value(flash) = $50

  • Expected value: resample camera price, take avg.

Patrick MacAlpine

slide-27
SLIDE 27

Spectrum licenses

  • Worth a lot
  • But how much to whom?

Patrick MacAlpine

slide-28
SLIDE 28

Spectrum licenses

  • Worth a lot
  • But how much to whom?
  • Used to be assigned

Patrick MacAlpine

slide-29
SLIDE 29

Spectrum licenses

  • Worth a lot
  • But how much to whom?
  • Used to be assigned

− took too long

Patrick MacAlpine

slide-30
SLIDE 30

Spectrum licenses

  • Worth a lot
  • But how much to whom?
  • Used to be assigned

− took too long

  • Switched to lotteries

Patrick MacAlpine

slide-31
SLIDE 31

Spectrum licenses

  • Worth a lot
  • But how much to whom?
  • Used to be assigned

− took too long

  • Switched to lotteries

− too random − clear that lots of value given away

Patrick MacAlpine

slide-32
SLIDE 32

Spectrum licenses

  • Worth a lot
  • But how much to whom?
  • Used to be assigned

− took too long

  • Switched to lotteries

− too random − clear that lots of value given away So decided to auction

Patrick MacAlpine

slide-33
SLIDE 33

Goals of mechanism

  • Efficient allocation (assign to whom it’s worth the most)
  • Promote deployment of new technologies
  • Prevent monopoly (or close)
  • Get some licenses to designated companies
  • No political embarrassments

Patrick MacAlpine

slide-34
SLIDE 34

Goals of mechanism

  • Efficient allocation (assign to whom it’s worth the most)
  • Promote deployment of new technologies
  • Prevent monopoly (or close)
  • Get some licenses to designated companies
  • No political embarrassments

Revenue an afterthought (but important in end)

Patrick MacAlpine

slide-35
SLIDE 35

Choices

  • Which basic auction format?

Patrick MacAlpine

slide-36
SLIDE 36

Choices

  • Which basic auction format?
  • Sequential or simultaneous auctions?

Patrick MacAlpine

slide-37
SLIDE 37

Choices

  • Which basic auction format?
  • Sequential or simultaneous auctions?
  • Combinatorial bids allowed?

Patrick MacAlpine

slide-38
SLIDE 38

Choices

  • Which basic auction format?
  • Sequential or simultaneous auctions?
  • Combinatorial bids allowed?
  • How to encourage designated companies?

Patrick MacAlpine

slide-39
SLIDE 39

Choices

  • Which basic auction format?
  • Sequential or simultaneous auctions?
  • Combinatorial bids allowed?
  • How to encourage designated companies?
  • Up front payments or royalties?

Patrick MacAlpine

slide-40
SLIDE 40

Choices

  • Which basic auction format?
  • Sequential or simultaneous auctions?
  • Combinatorial bids allowed?
  • How to encourage designated companies?
  • Up front payments or royalties?
  • Reserve prices?

Patrick MacAlpine

slide-41
SLIDE 41

Choices

  • Which basic auction format?
  • Sequential or simultaneous auctions?
  • Combinatorial bids allowed?
  • How to encourage designated companies?
  • Up front payments or royalties?
  • Reserve prices?
  • How much information public?

Patrick MacAlpine

slide-42
SLIDE 42

Problems from New Zealand and Australia

Second price, sealed bid

Patrick MacAlpine

slide-43
SLIDE 43

Problems from New Zealand and Australia

Second price, sealed bid

  • High bidder’s willingness to pay is public
  • No reserve prices
  • No penalties for default, so many meaningless high bids

Patrick MacAlpine

slide-44
SLIDE 44

Problems from New Zealand and Australia

Second price, sealed bid

  • High bidder’s willingness to pay is public
  • No reserve prices
  • No penalties for default, so many meaningless high bids

Any oversight in auction design can have harmful repercussions, as bidders can be counted on to seek ways to outfox the mechanism.

Patrick MacAlpine

slide-45
SLIDE 45

License interactions

  • Complementarities: good to be able to offer roaming

capabilities

Patrick MacAlpine

slide-46
SLIDE 46

License interactions

  • Complementarities: good to be able to offer roaming

capabilities

  • Substitutability: several licenses in the same region

Patrick MacAlpine

slide-47
SLIDE 47

License interactions

  • Complementarities: good to be able to offer roaming

capabilities

  • Substitutability: several licenses in the same region
  • Need

to be flexible to allow bidders to create aggregations

Patrick MacAlpine

slide-48
SLIDE 48

License interactions

  • Complementarities: good to be able to offer roaming

capabilities

  • Substitutability: several licenses in the same region
  • Need

to be flexible to allow bidders to create aggregations

  • Secondary market might allow for some corrections

− Likely to be thin − High transaction costs

Patrick MacAlpine

slide-49
SLIDE 49

Limits of Theory

Patrick MacAlpine

slide-50
SLIDE 50

Limits of Theory

  • Identify variables, but not relative magnitudes

Patrick MacAlpine

slide-51
SLIDE 51

Limits of Theory

  • Identify variables, but not relative magnitudes

− When there are conflicting effects, can’t tell which will dominate

Patrick MacAlpine

slide-52
SLIDE 52

Limits of Theory

  • Identify variables, but not relative magnitudes

− When there are conflicting effects, can’t tell which will dominate

  • Ignores transaction costs of implementing policies

Patrick MacAlpine

slide-53
SLIDE 53

Limits of Theory

  • Identify variables, but not relative magnitudes

− When there are conflicting effects, can’t tell which will dominate

  • Ignores transaction costs of implementing policies
  • May depend on unknown information

− e.g. bidder valuations

Patrick MacAlpine

slide-54
SLIDE 54

Limits of Theory

  • Identify variables, but not relative magnitudes

− When there are conflicting effects, can’t tell which will dominate

  • Ignores transaction costs of implementing policies
  • May depend on unknown information

− e.g. bidder valuations

  • Doesn’t scale to complexity of spectrum auctions

Patrick MacAlpine

slide-55
SLIDE 55

Limits of Theory

  • Identify variables, but not relative magnitudes

− When there are conflicting effects, can’t tell which will dominate

  • Ignores transaction costs of implementing policies
  • May depend on unknown information

− e.g. bidder valuations

  • Doesn’t scale to complexity of spectrum auctions

Used laboratory experiments too

Patrick MacAlpine

slide-56
SLIDE 56

Open vs. Sealed Bid

  • Open increases information, reducing winner’s curse

Patrick MacAlpine

slide-57
SLIDE 57

Open vs. Sealed Bid

  • Open increases information, reducing winner’s curse

− Leads to higher bids

Patrick MacAlpine

slide-58
SLIDE 58

Open vs. Sealed Bid

  • Open increases information, reducing winner’s curse

− Leads to higher bids

  • But. . .

− Risk aversion leads to higher bids in sealed bid auctions − Sealed bid auctions deter colusion

Patrick MacAlpine

slide-59
SLIDE 59

Open vs. Sealed Bid

  • Open increases information, reducing winner’s curse

− Leads to higher bids

  • But. . .

− Risk aversion leads to higher bids in sealed bid auctions − Sealed bid auctions deter colusion

  • Decided former outweighed latter
  • Went with announcing bids, but not the bidders

Patrick MacAlpine

slide-60
SLIDE 60

Open vs. Sealed Bid

  • Open increases information, reducing winner’s curse

− Leads to higher bids

  • But. . .

− Risk aversion leads to higher bids in sealed bid auctions − Sealed bid auctions deter colusion

  • Decided former outweighed latter
  • Went with announcing bids, but not the bidders

− Circumvented!

Patrick MacAlpine

slide-61
SLIDE 61

Simultaneous vs. Sequential

  • Sequential prevents backup strategies for aggregation
  • Sequential also allows for budget stretching

Patrick MacAlpine

slide-62
SLIDE 62

Simultaneous vs. Sequential

  • Sequential prevents backup strategies for aggregation
  • Sequential also allows for budget stretching
  • Simultaneous needs a stopping rule

− Closing one by one is effectively sequential − Keeping all open until all close encourages sniping

Patrick MacAlpine

slide-63
SLIDE 63

Simultaneous vs. Sequential

  • Sequential prevents backup strategies for aggregation
  • Sequential also allows for budget stretching
  • Simultaneous needs a stopping rule

− Closing one by one is effectively sequential − Keeping all open until all close encourages sniping

  • Stopping rule should:

− End auction quickly − Close licenses almost simultaneously − be simple and understandable

Patrick MacAlpine

slide-64
SLIDE 64

Simultaneous vs. Sequential

  • Sequential prevents backup strategies for aggregation
  • Sequential also allows for budget stretching
  • Simultaneous needs a stopping rule

− Closing one by one is effectively sequential − Keeping all open until all close encourages sniping

  • Stopping rule should:

− End auction quickly − Close licenses almost simultaneously − be simple and understandable Went with activity rules

Patrick MacAlpine

slide-65
SLIDE 65

Combinatorial Bids

  • Nationwide

bidding could decrease efficiency and revenue

Patrick MacAlpine

slide-66
SLIDE 66

Combinatorial Bids

  • Nationwide

bidding could decrease efficiency and revenue

  • Full combinatorial bidding too complex

− Winner determination problem − Active research area

Patrick MacAlpine

slide-67
SLIDE 67

Aiding Designated Bidders

  • Give them a discount

Patrick MacAlpine

slide-68
SLIDE 68

Aiding Designated Bidders

  • Give them a discount
  • Circumvented!

Patrick MacAlpine

slide-69
SLIDE 69

Royalties vs. Up-front Payments

  • Royalties decrease risk, increase bids

Patrick MacAlpine

slide-70
SLIDE 70

Royalties vs. Up-front Payments

  • Royalties decrease risk, increase bids
  • But royalties discourage post-auction innovation

Patrick MacAlpine

slide-71
SLIDE 71

Royalties vs. Up-front Payments

  • Royalties decrease risk, increase bids
  • But royalties discourage post-auction innovation
  • Decided against

Patrick MacAlpine

slide-72
SLIDE 72

Reserve Prices

  • Not necessary in such a competitive market
  • Did include withdrawal penalties

Patrick MacAlpine

slide-73
SLIDE 73

Results

  • Big successes

− Lots of bidders − Lots of revenue

Patrick MacAlpine

slide-74
SLIDE 74

Results

  • Big successes

− Lots of bidders − Lots of revenue

  • Also some problems

− Strategic Demand Reduction

Patrick MacAlpine

slide-75
SLIDE 75

Results

  • Big successes

− Lots of bidders − Lots of revenue

  • Also some problems

− Strategic Demand Reduction

  • Incremental design changes

− New problems always arise − Bidders indeed find ways to circumvent mechanisms

Patrick MacAlpine

slide-76
SLIDE 76

Results

  • Big successes

− Lots of bidders − Lots of revenue

  • Also some problems

− Strategic Demand Reduction

  • Incremental design changes

− New problems always arise − Bidders indeed find ways to circumvent mechanisms

  • Lessons to be learned via agent-based experiments

Patrick MacAlpine

slide-77
SLIDE 77

Discussion

  • How could you fix the aspects that were circumvented?

Patrick MacAlpine

slide-78
SLIDE 78

Discussion

  • How could you fix the aspects that were circumvented?
  • Could you design a better auction mechanism?

Patrick MacAlpine

slide-79
SLIDE 79

Discussion

  • How could you fix the aspects that were circumvented?
  • Could you design a better auction mechanism?
  • Best bidding strategies?

Patrick MacAlpine

slide-80
SLIDE 80

Discussion

  • How could you fix the aspects that were circumvented?
  • Could you design a better auction mechanism?
  • Best bidding strategies?
  • Use of agents in FCC spectrum auction?

Patrick MacAlpine

slide-81
SLIDE 81

Discussion

  • How could you fix the aspects that were circumvented?
  • Could you design a better auction mechanism?
  • Best bidding strategies?
  • Use of agents in FCC spectrum auction?
  • Need to know entire agent preference...

Patrick MacAlpine

slide-82
SLIDE 82

Discussion

  • How could you fix the aspects that were circumvented?
  • Could you design a better auction mechanism?
  • Best bidding strategies?
  • Use of agents in FCC spectrum auction?
  • Need to know entire agent preference...
  • Multiple item bidding in RoboCup?

Patrick MacAlpine

slide-83
SLIDE 83

FCC Spectrum Auction #35

  • 422 licences in 195 markets (cities)

− 80 bidders spent $8 billion − ran Dec 12 - Jan 26 2001 − licence is a 10 or 15 mhz spectrum chunk

  • Run in rounds

− bid on each licence you want each round − simultaneous; break ties by arrival time − current winner and all bids are known

  • Allowable bids: 1 to 9 bid increments

− 1 bid incr is 10% – 20% of current price

  • Other complex rules

Patrick MacAlpine