A comparison-based gradient estimation algorithm for sequential multi-issue resource trading between two greedily rational agents guarantees strict utility improvements per accepted trade and asymptotic convergence to the Pareto front under mild assumptions.
Since we assume that the responding agent’s utility function is smooth and the offers have a bounded magnitude, the change in the responding agent’s gradient direction is bounded
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.MA 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Sequential Resource Trading Using Comparison-Based Gradient Estimation
A comparison-based gradient estimation algorithm for sequential multi-issue resource trading between two greedily rational agents guarantees strict utility improvements per accepted trade and asymptotic convergence to the Pareto front under mild assumptions.