Best of Both Worlds: Ex-Ante and Ex-Post Fairness in Resource - - PowerPoint PPT Presentation

best of both worlds
SMART_READER_LITE
LIVE PREVIEW

Best of Both Worlds: Ex-Ante and Ex-Post Fairness in Resource - - PowerPoint PPT Presentation

Best of Both Worlds: Ex-Ante and Ex-Post Fairness in Resource Allocation Nisarg Shah University of Toronto Joint work with Rupert Freeman (MSR NYC, now at UVA) & Rohit Vaish (RPI) Implementing Fractional Allo Im llocations 1 1


slide-1
SLIDE 1

Best of Both Worlds:

Ex-Ante and Ex-Post Fairness in Resource Allocation

Joint work with Rupert Freeman (MSR NYC, now at UVA) & Rohit Vaish (RPI)

Nisarg Shah

University of Toronto

slide-2
SLIDE 2

Im Implementing Fractional Allo llocations

EC 2020 - Best of Both Worlds 2 Nisarg Shah

=

1 3 × ∅

Ex-ante Fair Ex-post Unfair Ex-post Unfair Ex-post Unfair

∅ ∅ ∅ ∅ +

1 3 ×

+

1 3 ×

slide-3
SLIDE 3

Im Implementing Fractional Allo llocations

EC 2020 - Best of Both Worlds 3 Nisarg Shah

=

1 3 ×

Ex-ante Fair Ex-post Fair Ex-post Fair Ex-post Fair

+

1 3 ×

+

1 3 ×

slide-4
SLIDE 4

Sele lected Results

EC 2020 - Best of Both Worlds 4 Nisarg Shah

Can we always achieve ex-ante fairness by randomizing over ex-post fair allocations?

(Efficiently) possible: ex-ante EF + ex-post EF1 (Efficiently) possible: ex-ante GF + ex-post Prop1 + ex-post EF1

1

Impossible: ex-ante Prop + ex-post EF1 + ex-post fPO Extensions to division of chores