fairness efficiency tradeoffs in dynamic

Fairness-Efficiency Tradeoffs in Dynamic Fair Division David Zeng, - PowerPoint PPT Presentation

Fairness-Efficiency Tradeoffs in Dynamic Fair Division David Zeng, Alex Psomas items arrive online, agents Agent has value [0,1] for item that we learn when the item arrives Item must be allocated


  1. Fairness-Efficiency Tradeoffs in Dynamic Fair Division David Zeng, Alex Psomas β€’ π‘ˆ items arrive online, π‘œ agents β€’ Agent 𝑗 has value 𝑀 𝑗𝑒 ∈ [0,1] for item 𝑒 that we learn when the item arrives β€’ Item must be allocated immediately and irrevocably β€’ After π‘ˆ rounds, we will have some allocation 𝐡 = (𝐡 1 , … , 𝐡 π‘œ ) β€’ Additive valuations: 𝑀 𝑗 𝐡 π‘˜ = Οƒ 𝑕 𝑒 ∈𝐡 π‘˜ 𝑀 𝑗𝑒 β€’ Ideally, allocation is both fair and efficient

  2. Adversary Model and Results 1 No π‘œ + 𝜁 -Pareto efficient and sublinear envy allocation algorithm Non-adaptive Adaptive 𝑀 𝑒 ~𝐸 Τ¦ 𝑀 𝑗𝑒 ~𝐸 𝑀 𝑗𝑒 ~𝐸 𝑗 adversary adversary Fairness and efficiency Fairness and efficiency compatible incompatible ex-post Pareto efficient and pairwise (EF1 or EF w.h.p.) allocation algorithm

  3. Algorithm for Correlated Agents ( Τ¦ 𝑀 𝑒 ~𝐸 ) β€’ Reduce finding a β€’ online ex-post Pareto efficient and (EF1 or EF w.h.p.) allocation algorithm to finding a β€’ offline Pareto efficient and CISEF fractional allocation β€’ Algorithm sketch Given online problem and distribution 𝐸 with support 𝛿 1 , … , 𝛿 𝑛 , use the β€’ support of 𝐸 as the items for offline problem, scaling by the probabilities. Use the fractional allocation π‘Œ to guide our allocation in the online problem. β€’ If π‘Œ 𝑗𝑙 = 0.4 , if the item arriving at time 𝑒 has type 𝛿 𝑙 , allocate the item to agent 𝑗 with β€’ probability 0.4 Treat cliques as one combined agent when doing randomized allocation β€’ When item is allocated to the clique, give to unhappiest agent in clique β€’

  4. CISEF Clique Identical Strongly Envy-Free β€’ CISEF 1 2 Either agent 𝑗 strictly prefers her own β€’ bundle to the bundle of agent π‘˜ Or 𝑗 and π‘˜ have identical allocations β€’ and the same value (up to a scaling factor) for all the items that are allocated to either of them β€’ How to find CISEF and Pareto Identical allocations efficient allocation? and valuations 3 Start with solution to Eisenberg-Gale β€’ convex program Strongly envy-free If agent 𝑗 is indifferent to agent π‘˜ , β€’ (carefully) move items from π‘˜ to 𝑗 to create strong envy-free edges

Recommend


More recommend