#### DMCA

## Selective call out and real time bidding (2010)

Venue: | In Internet and Network Economics |

Citations: | 7 - 4 self |

### Citations

5219 | Convex Analysis
- Rockafellar
- 1970
(Show Context)
Citation Context ... framework to the stochastic variants of the Adwords problem [9, 27]. But while the similarity implies that Lagrangian decoupling techniques for separable convex optimization pioneered by Rockafellar =-=[25]-=- apply, the different possible objectives of the call-out framework are not convex. In fact, in the case of optimizing revenue in GSP (with reserve) or in posted price mechanisms, the objective is not... |

1563 |
Optimal auction design
- Myerson
- 1981
(Show Context)
Citation Context ...ernal) then we would only call out the winners (assuming multiple slots) of the auction, which is the path taken by the Adwords problem. The call-out framework is similar to Bayesian mechanism design =-=[21]-=-. This has some fairly broad conceptual implications. First, is the notion of “adaptivity gap”, where a policy is allowed to react to realization of the random variables. The analysis of adaptivity ga... |

143 | Adwords and generalized on-line matching
- Mehta, Saberi, et al.
- 2005
(Show Context)
Citation Context ...ed in Section 1.1. The online allocation aspect, with a view that the call out constraints acts as budgets, is reminiscent of the online ad allocation framework for search ads, or the Adwords problem =-=[19, 5, 7]-=-, and its stochastic variants [9, 27]. However the call out framework is significantly different, which we discuss below. The Adwords problem is posed in the deterministic setting where the expected r... |

125 |
Computer Networks, 3rd Edition
- Tanenbaum
- 1996
(Show Context)
Citation Context ...ade on contiguous subset of impressions. This misses the original goal that the ad network would receive the impressions at a “smooth” rate. A common model used for behavior is the token bucket model =-=[26]-=-. A token bucket has two parameters, bucket size σ and token generation rate ρ. The tokens represent sending rights, and the bucket size is the maximum number of tokes we can store. The tokens are gen... |

112 | Maximizing a submodular set function subject to a matroid constraint (extended abstract
- Calinescu, Chekuri, et al.
- 2007
(Show Context)
Citation Context ...echanisms, the objective is not submodular for all prices (as in welfare maximization). Submodular maximization with linear constraints has been studied, and while good approximation algorithms exist =-=[6, 17]-=-, they are inherently offline – the key aspect of call out optimization is that the decision has to be made in an online fashion. The same is true for sequential posted price mechanisms analyzed in [8... |

85 | Multi-Armed Bandits in Metric Spaces
- Kleinberg, Slivkins, et al.
- 2008
(Show Context)
Citation Context ...roblem. We note that the bandwidth-like constraints (where the constraint is on a parameter different than the obtained value, as is the case for call-outs) has not studied in the bandit setting (see =-=[16, 23]-=-) because the horizon is constrained. Finally, bidding and inventory optimization problems studied in the context of ad exchanges [11, 10, 20], are not related to the call out optimization problems. 2... |

71 | Online primal-dual algorithms for maximizing ad-auctions revenue - Buchbinder, Jain, et al. - 2007 |

68 | The adwords problem: Online keyword matching with budgeted bidders under random permutations
- Devanur, Hayes
- 2009
(Show Context)
Citation Context ... aspect, with a view that the call out constraints acts as budgets, is reminiscent of the online ad allocation framework for search ads, or the Adwords problem [19, 5, 7], and its stochastic variants =-=[9, 27]-=-. However the call out framework is significantly different, which we discuss below. The Adwords problem is posed in the deterministic setting where the expected revenue is treated as a known determin... |

66 | Allocating bandwidth for bursty connections
- Kleinberg, Rabani, et al.
- 2000
(Show Context)
Citation Context ...a contrast to the stochastic results, the adversarial order setting is discussed in Appendix A. Other Related Work: A combination of stochastic and online components appear in many different settings =-=[14, 15, 2, 13]-=- which are not immediately relevant to the call-out problem. We note that the bandwidth-like constraints (where the constraint is on a parameter different than the obtained value, as is the case for c... |

62 |
Properties of probability distributions with monotone hazard rate
- Barlow, Marshall, et al.
- 1963
(Show Context)
Citation Context ...enue that is O(1) factor of optimal welfare, when all bid distributions satisfy the MHR property. This is a common distributional assumption in economic theory, and is satisfied by many distributions =-=[3]-=-. This result is in the same spirit as (but immediately incomparable to) the result in [4], which relates the optimum sequential posted price revenue to the optimal welfare under the same assumptions.... |

49 | R.: A knapsack secretary problem with applications. In: APPROX-RANDOM
- Babaioff, Immorlica, et al.
- 2007
(Show Context)
Citation Context ...a contrast to the stochastic results, the adversarial order setting is discussed in Appendix A. Other Related Work: A combination of stochastic and online components appear in many different settings =-=[14, 15, 2, 13]-=- which are not immediately relevant to the call-out problem. We note that the bandwidth-like constraints (where the constraint is on a parameter different than the obtained value, as is the case for c... |

49 | Maximizing submodular set functions subject to multiple linear constraints
- Kulik, Shachnai, et al.
(Show Context)
Citation Context ...echanisms, the objective is not submodular for all prices (as in welfare maximization). Submodular maximization with linear constraints has been studied, and while good approximation algorithms exist =-=[6, 17]-=-, they are inherently offline – the key aspect of call out optimization is that the decision has to be made in an online fashion. The same is true for sequential posted price mechanisms analyzed in [8... |

34 | On the approximability of budgeted allocations and improved lower bounds for submodular welfare maximization and GAP
- Chakrabarty, Goel
(Show Context)
Citation Context ...ed in Section 1.1. The online allocation aspect, with a view that the call out constraints acts as budgets, is reminiscent of the online ad allocation framework for search ads, or the Adwords problem =-=[19, 5, 7]-=-, and its stochastic variants [9, 27]. However the call out framework is significantly different, which we discuss below. The Adwords problem is posed in the deterministic setting where the expected r... |

32 |
Ad exchanges: Research issues
- Muthukrishnan
- 2009
(Show Context)
Citation Context ...ase for call-outs) has not studied in the bandit setting (see [16, 23]) because the horizon is constrained. Finally, bidding and inventory optimization problems studied in the context of ad exchanges =-=[11, 10, 20]-=-, are not related to the call out optimization problems. 2 Preliminaries Consider a maximizing a “separable” linear program (LP) L defined on Q global constraints with right hand side bi, such that th... |

28 | Stochastic on-line knapsack problems - Marchetti-Spaccamela, Vercellis - 1995 |

25 | Budget constrained auctions with heterogeneous items. CoRR abs/0907.4166
- Goel, Gollapudi, et al.
- 2009
(Show Context)
Citation Context ...7], they are inherently offline – the key aspect of call out optimization is that the decision has to be made in an online fashion. The same is true for sequential posted price mechanisms analyzed in =-=[8, 4]-=- (albeit with more general matroid setting), the posted prices in the call out setting need to be announced in parallel (and the eventual allocation is sequential). Other than formulating the call out... |

25 | Approximate quantiles and the order of the stream
- Guha, McGregor
- 2006
(Show Context)
Citation Context ...grangians derived from a small number of samples (suitably scaled) can be used to solve the overall LP. The weighted sampling reduces to the prefix of the input if the Ljs arrive in random order (see =-=[12]-=-). This was extended to convex programs in [27]. The number of sample bound requires several (easy) Lipschitz type properties: 1. The optimum value of L is at least δ > 0 and the optimum solution of L... |

25 | Handling advertisements of unknown quality in search advertising
- Pandey, Olston
- 2006
(Show Context)
Citation Context ...roblem. We note that the bandwidth-like constraints (where the constraint is on a parameter different than the obtained value, as is the case for call-outs) has not studied in the bandit setting (see =-=[16, 23]-=-) because the horizon is constrained. Finally, bidding and inventory optimization problems studied in the context of ad exchanges [11, 10, 20], are not related to the call out optimization problems. 2... |

22 | Sequential posted pricing and multi-parameter mechanism design
- Chawla, Hartline, et al.
(Show Context)
Citation Context ...7], they are inherently offline – the key aspect of call out optimization is that the decision has to be made in an online fashion. The same is true for sequential posted price mechanisms analyzed in =-=[8, 4]-=- (albeit with more general matroid setting), the posted prices in the call out setting need to be announced in parallel (and the eventual allocation is sequential). Other than formulating the call out... |

21 | S.: Bidding for representative allocations for display advertising
- Ghosh, McAfee, et al.
(Show Context)
Citation Context ...ase for call-outs) has not studied in the bandit setting (see [16, 23]) because the horizon is constrained. Finally, bidding and inventory optimization problems studied in the context of ad exchanges =-=[11, 10, 20]-=-, are not related to the call out optimization problems. 2 Preliminaries Consider a maximizing a “separable” linear program (LP) L defined on Q global constraints with right hand side bi, such that th... |

20 | Optimal online assignment with forecasts
- Vee, Vassilvitskii, et al.
- 2010
(Show Context)
Citation Context ... aspect, with a view that the call out constraints acts as budgets, is reminiscent of the online ad allocation framework for search ads, or the Adwords problem [19, 5, 7], and its stochastic variants =-=[9, 27]-=-. However the call out framework is significantly different, which we discuss below. The Adwords problem is posed in the deterministic setting where the expected revenue is treated as a known determin... |

13 |
Adaptive bidding for display advertising
- Ghosh, Rubinstein, et al.
(Show Context)
Citation Context ...ase for call-outs) has not studied in the bandit setting (see [16, 23]) because the horizon is constrained. Finally, bidding and inventory optimization problems studied in the context of ad exchanges =-=[11, 10, 20]-=-, are not related to the call out optimization problems. 2 Preliminaries Consider a maximizing a “separable” linear program (LP) L defined on Q global constraints with right hand side bi, such that th... |

7 |
A multiple-choice secretary problem with applications to online auctions
- Kleinberg
- 2005
(Show Context)
Citation Context ...a contrast to the stochastic results, the adversarial order setting is discussed in Appendix A. Other Related Work: A combination of stochastic and online components appear in many different settings =-=[14, 15, 2, 13]-=- which are not immediately relevant to the call-out problem. We note that the bandwidth-like constraints (where the constraint is on a parameter different than the obtained value, as is the case for c... |

4 | On revenue maximization in second-price ad auctions
- Azar, Birnbaum, et al.
- 2009
(Show Context)
Citation Context ...serve price, before the bids are solicited as in [21]. For known deterministic bids, reserve prices can be made equal to the bid, and are not useful. Strong lower bounds hold for GSP without reserves =-=[1]-=-. Third, the notion of a comparison class in case of call out optimization framework requires more care. In the setting of these large exchanges, a comparison class with full foreknowledge of all info... |

3 | Adaptive Uncertainty Resolution in Bayesian Combinatorial Optimization Problems
- Guha, Munagala
- 2008
(Show Context)
Citation Context |