Price optimization with robust learning constraint
A valuation optimization method includes generating offeree decision information (buyer decision information, by way of illustrative example) by presenting a sequence of mechanisms to a sequence of offerees wherein the mechanisms comprise menus of transaction offers (sale offer menus, by way of illustrative example). Actual transactions (sale transactions, by way of illustrative example) are conducted responsive to acceptances of transaction offers by buyers. At a selected time in the generating, an offeree valuation distribution belief and the current mechanism are updated optimize an offeree's utility. The offeree's utility comprises an offeree's utility function constrained by a robust learning constraint computed based on a local differential of an earlier offeree's utility function with respect to the earlier offeree's valuation.
1. A method comprising: generating buyer decision information by presenting a sequence of sale offer menus to a sequence of buyers and conducting actual sale transactions based on sale offers accepted by buyers; and at a selected time in the generating, updating a buyer valuation distribution belief and the current sale offer menu of the sequence of sale offer menus to optimize a buyer's utility comprising a buyer's utility function constrained by a robust learning constraint that is computed based on a local differential of an earlier buyers utility function with respect to the earlier buyer's valuation, the robust learning constraint bounding a rate of change of the later buyer's utility based on the local curvature of the earlier buyer's utility function at the earlier buyer's valuation; wherein the generating and the updating operations are performed by a digital processor.
2. The method of claim 1 , wherein the updating is performed occasionally during the generating to iteratively optimize the buyer valuation distribution belief and the sale offer menu.
3. The method of claim 1 , wherein the robust learning constraint bounds the rate of change of the later buyer's utility based on a second derivative with respect to valuation of the earlier buyer's utility function at the earlier buyer's valuation.
4. The method of claim 1 , wherein the robust learning constraint is parameterized by a tuning parameter whose value adjusts the robust learning constraint.
5. The method of claim 1 , wherein the updating comprises:
solving a linear program embodying the buyer's utility comprising the buyer's utility function constrained by the robust learning constraint.
6. The method of claim 1 , wherein the updating comprises:
solving a semi-definite program (SDP) embodying the buyer's utility comprising the buyer's utility function constrained by the robust learning constraint.
7. An apparatus comprising: a digital processor configured to perform a method including: generating buyer decision information by presenting a sequence of sale offer menus to a sequence of buyers and conducting actual sale transactions based on sale offers accepted by buyers; and at a selected time in the generating, updating a buyer valuation distribution belief and the current sale offer menu of the sequence of sale offer menus to optimize a buyer's utility comprising a buyer's utility function constrained by a robust learning constraint that is computed based on a local differential of an earlier buyer's utility function with respect to the earlier buyer's valuation, the robust learning constraint bounding a rate of change of the later buyer's utility based on the local curvature of the earlier buyer's utility function at the earlier buyer's valuation.
8. A non-transitory storage medium storing instructions executable on a digital processor to perform a method comprising: generating offeree decision information by presenting a sequence of mechanisms to a sequence of offerees wherein the mechanisms comprise menus of transaction offers and conducting actual transactions responsive to acceptances of transaction offers by offerees; and at a selected time in the generating, updating an offeree valuation distribution belief and the current mechanism to optimize an offeree's utility comprising an offeree's utility function constrained by a robust learning constraint computed based on a local differential of an earlier offeree's utility function with respect to the earlier offeree's valuation, the robust learning constraint bounding a rate of change of the later offeree's utility based on the local curvature of the earlier offeree's utility function at the earlier offeree's valuation.
9. The non-transitory storage medium of claim 8 , wherein the sequence of offerees comprises one of (i) a sequence of buyers and (ii) a sequence of sellers or suppliers.
10. The non-transitory storage medium of claim 8 , wherein the updating is performed occasionally during the generating to iteratively optimize the offeree valuation distribution belief and the mechanism.
11. The non-transitory storage medium of claim 8 , wherein the robust learning constraint bounds the rate of change of the later offeree's utility based on a second derivative with respect to valuation of the earlier offeree's utility function at the earlier offeree's valuation.
12. The non-transitory storage medium of claim 8 , wherein the robust learning constraint is parameterized by a tuning parameter whose value adjusts the robust learning constraint.
13. The non-transitory storage medium of claim 8 , wherein the updating comprises: solving a linear program embodying the offeree's utility comprising the offeree's utility function constrained by the robust learning constraint.
14. The non-transitory storage medium of claim 8 , wherein the updating comprises: solving a semi-definite program (SDP) embodying the offeree's utility comprising the offeree's utility function constrained by the robust learning constraint.